docs: fix dialect.py description — AAAK is lossy, not "30x lossless"

[moltate]

What this fixes

The File Reference table describes dialect.py as:

AAAK compression — entity codes, emotion markers, 30x lossless ratio

This contradicts the maintainers' own April 7 correction, which states:

"30x lossless compression was overstated. AAAK is a lossy abbreviation
system (entity codes, sentence truncation)."

Change made

Updated the description to:

AAAK compression — entity codes, emotion markers, lossy abbreviation (experimental)

This aligns with the honest status already documented in the AAAK Dialect
section of the README.

No code changes

README documentation only.

[milla-jovovich]

Thanks for flagging this — we actually caught it ourselves and fixed it in PR #835 last week. The file
reference table now says 'AAAK index format for closet pointers' on both main and develop. We also added 42 TDD tests that verify every README claim against the actual code, including this one, so it can't drift again. Closing as already resolved. Thank you from Milla and Lu✨v

[moltate]

Thanks for the quick update and the detailed explanation! Glad to hear it was already fixed in #835 and that you added those TDD tests — that's a solid improvement. Appreciate the review from you and Lu. I'll check out the changes.

Best,
Ankit