Recognise gpt-5.1 model identifier (Fixes #464)

[jbbqqf]

Summary

Fixes #464. encoding_for_model("gpt-5.1") currently raises KeyError
because the prefix table only contains "gpt-5-", and "gpt-5.1" does
not start with "gpt-5-" (the literal hyphen mismatches the dot — this
is the same shape that caused #464 to be filed).

This change adds "gpt-5.1" (exact) and "gpt-5.1-" (versioned
variant) to the o200k_base mappings, alongside the existing "gpt-5" /
"gpt-5-" pair. Per OpenAI's GPT-5.1 migration note
(https://platform.openai.com/docs/guides/latest-model#migrating-from-other-models-to-gpt-5-1)
the model is documented as a drop-in replacement for GPT-5, so the
encoding choice mirrors GPT-5's.

A short code comment in MODEL_PREFIX_TO_ENCODING notes the
ordering rationale (purely cosmetic given startswith semantics, but
makes the intent explicit for the next person editing the table).

Reproduce BEFORE/AFTER yourself (copy-paste)

git clone https://github.com/openai/tiktoken.git /tmp/tt-464 && cd /tmp/tt-464
pip install -e . pytest

# BEFORE — on origin/main, gpt-5.1 raises KeyError
git checkout main
python -c "
import tiktoken
print(tiktoken.encoding_for_model('gpt-5').name)
try:
    print(tiktoken.encoding_for_model('gpt-5.1').name)
except KeyError as e:
    print('FAIL:', e)
"
# Expected: gpt-5 -> o200k_base; gpt-5.1 -> KeyError

# AFTER — on this branch, gpt-5.1 maps to o200k_base
git fetch https://github.com/jbbqqf/tiktoken.git feat/464-gpt-5.1-encoding
git checkout FETCH_HEAD
pip install -e .
python -c "
import tiktoken
for m in ['gpt-5', 'gpt-5.1', 'gpt-5.1-2025-11', 'gpt-5-mini']:
    print(m, '->', tiktoken.encoding_for_model(m).name)
"
# Expected: all four map to o200k_base

What I ran locally

$ pytest tests/test_misc.py::test_encoding_for_model -v
tests/test_misc.py::test_encoding_for_model PASSED

$ pytest
============== 33 passed in 1.80s ==============

Verified the new test fails on origin/main (before the tiktoken/model.py
patch is applied) with KeyError: 'Could not automatically map gpt-5.1 to a tokeniser', and passes after the patch.

Edge cases

Input	Behavior before	Behavior after
"gpt-5"	o200k_base (exact match)	unchanged
"gpt-5-mini"	o200k_base via gpt-5- prefix	unchanged
"gpt-5.1"	KeyError	o200k_base (exact match)
"gpt-5.1-2025-11"	KeyError	o200k_base via gpt-5.1- prefix
"gpt-5.2" (hypothetical future)	KeyError	still KeyError (intentional — wait for explicit OpenAI doc)
Any non-gpt-5 model	unchanged	unchanged

The change is additive (two new dict entries plus a comment); existing
mappings are untouched.

---

PR drafted with assistance from Claude Code (Anthropic). The change
was reviewed manually against tiktoken/model.py and the OpenAI
migration doc linked above. The reproducer block above is the one I
used during development; reviewers can paste it verbatim.