To address a couple of points in more detail...
Why it doesn't have to be a conlang
- it's entirely natural to have a dominant word-form as biblical hebrew does.
- given the number of consonants in the language, it's possible to perfectly express anything you'd want to express with three-consonant roots. People don't just go around adding syllables to things willy-nilly.
- the great variability in the length of English words is due to our extensive derivational processes, but more importantly due to borrowing, which creates new 'roots' that were derived forms in a parent language.
To quantify that...
that - is - right - yes - to - a - of - points - in - more - why - it - have - be - word - form - as - does - the - you'd - want - with - three - roots - don't - just - go -things - great - length - words - due - our - but - which - creates - new - were - forms
To which we can add the following, which are one-syllable roots with clear affixes added:
broad+ly - does+n't - give+n - vary+able+ity - add-ing
And a word that is an English affix added to a one-syllable 'root' that English itself doesn't have, but the parent language did:
Words that are ultimately borrowed from Latin derived forms with prepositional prefixes (or occasionally similar derivatives formed in French), where the root itself may not have survived into English, or may have been borrowed with an unrelated meaning:
ad-dress - de-tail - en-tire-ly - per-fect-ly - ex-press - ex-tens-ive - de-rive-d - pro-cess-es
Words that also contain obvious Latin case-marking inflections:
con-son-ant-s - con-son-ant - par-ent
Words that incorporate transparent Latin part-of-speech and/or diminutive derivational affixes no longer productive in English:
coup-le* - dom-in-ant - nat-ur-al - lang-uage* - de-riv-ation-al - im-port-ant-ly
(*not transparent in English, adminitedly, because of the French intermediary. Ultimately cop+ula ('little joining together') and lingua + aticum, so "tongue-ish (thing)")
Words where the Latin root is bisyllabic due to affix of unclear purpose, or possible borrowing from outside latin:
numb-er - people*
(*Latin poplus, possibly a borrowing from Etruscan)
One-syllable English root with an original affix of only borderline continued productivity:
One-syllable English root plus affixes, where the root has been lost:
Eng-l-ish (the root here survives in, eg, German, but is lost in English, other than in re-anylised 'hangnail' and borrowed 'anger')
Proper name borrowed from another language:
Word irregularly derived from a phrase consisting of single-syllable words:
Compound word formed from one-syllable roots and an affix:
Extremely recent compound word following a non-traditional abbreviating structure:
Borrowing from, ultimately, a Semitic triconsonantal root:
(originally from Phoenician G-B-L, used as the name of a city, misheard by Greeks (and given a Greek affix) as 'Byblos', the port where the Greeks got their papyrus from)
Borrowing from Greek, via French, where the affixes are now completly non-transparent:
(originally syn-lab-e, with the -le added as a mishearing along the way)
So as you can see, the inherited Germanic lexicon in English is extremely biased toward one-syllable roots. There are very few Germanic words in English that aren't one-syllable, or a one-syllable root with derivational or inflectional affixes, or a compound of such words. Most of the exceptions I think are words with a final -ow or -ough syllable - and those words were one-syllable in Old English ('borrow' is from the root borg-, and is still one-syllable in other Germanic languages, but English final -g after consonants has often vocalised). The English words that aren't based on one-syllable roots are overwhelmingly borrowings from other languages - most of them being borrowed from other IE languages that ALSO are extremely biased toward one-syllable roots, but where either the affix or the root has not made it into English (or has been borrowed with an unrelated meaning). This is because Indo-European languages overwhelmingly have inherited single-syllable roots. (Generally of the form CVC, sometimes with n/m, r, or l either before or after the vowel).
[English roots ending in a vowel typically descend from Germanic roots with a final consonant that has been vocalised. Eg "see" from sehw-, "sea" from "saiw-", "say" from "sag-" and so on]
Semitic languages, on the other hand, have inherited bisyllabic roots, but with simpler phonotactics, so generally just of the form CVCVC. Since they have experienced much less borrowing than English has - particularly when you're looking an ancient forms of the language - their roots show much less variation from this basic pattern than English roots do.
- having more variability in the nouns than the verbs probably just means that biblical hebrew makes more use of (potentially opaquely) derived and compounded nouns than of verbs, which is perfectly normal, or that it tends to borrow words as nouns rather than verbs (almost universal).
- sure, no written record really records speech. Either someone wrote down the Torah from scratch, or they based it on orally-transmitted texts, which tend to undergo changes to make them easier to remember (i.e. they tend to be poetry). But that doesn't make something a conlang. English isn't a conlang just because a lot of words and phrases come from Shakespeare.
- gematrial coincidences happen all the time through chance. Particularly when your calculations are all based on multiplying only three variables, which have a limited number of pre-defined values.
Why Hebrew can't be a conlang
- hebrew descends through regular sound changes from proto-Semitic, as do Arabic, Aramaic, Phoenician, Eblaite, etc.
- to create the appearance of that having happened, the 'creators' of hebrew would have had to a) recognise the existence of the Semitic family, b) be sufficiently knowledgeable about its members, c) reconstruct accurately the proto-Semitic family (or, at least, accurately enough that we can't spot the fake), and d) derive Hebrew from that language, in the process following principles of diachronic development that the rest of the world didn't discover until the 19th century, and then e) leave no trace of any of the vast linguistic scholarship that would have been required to have done this. Where, for example, are the manuscripts comparing the details of the inflections of Eblaite, Amorite and Phoenician verbs? And then f) they'd have to all abandon their own native language.
And g) they've have to have some reason for all of this.