feat: create a model from a tokenizer

[stephantul]

I found myself often creating random models from tokenizers, so I decided to add a function that just takes a tokenizer and then creates random vectors. This is useful, because the vocabulary creation/guards and checks around vocabulary were not available outside of distillation.

I added two functions:

1. random_init
2. init_from_tokenizer

The former takes a tokenizer name, the second takes an initialized tokenizer.

[greptile-apps]

Confidence Score: 4/5

[Medium impact] Adds new functions to initialize models from tokenizers.

The PR appears safe to merge, with two non-blocking improvements worth making.

_{Reviews (1) · Last reviewed commit: "mypy" · Reviewed by Greptile}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

Files with missing lines	Coverage Δ
model2vec/distill/__init__.py	100.00% <100.00%> (ø)
model2vec/distill/distillation.py	100.00% <100.00%> (+5.47%)	⬆️
model2vec/distill/inference.py	98.59% <100.00%> (+<0.01%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Pringled]

One smoll thing but LGTM! I think some of the new tests could become part of existing tests but fine either way