internal.h: shorten the mrb_str_valid_encoding_p comment - #7108

The comment on the declaration of mrb_str_valid_encoding_p, added in #7105,
restated the one on the definition in string.c: which flag the walk declines to
read on the way in, why reading it would make the answer depend on whether the
string had been measured before, and which flag a string that walks to one byte
per character is left with.

A caller reading the header needs the answer the function gives, not how it
arrives at one. Keep the three answers it has to know, which are the two strings
the walk never touches and the one it rejects, and leave the rest at the
definition, where the code that does the walking is.

/* Whether a string's bytes read as the encoding it is taken to have: FALSE for
   one holding a byte that stands for no character, TRUE for a binary string
   whatever its bytes are, and TRUE throughout on a non-UTF-8 build. See the
   definition in string.c for what it reads and what it leaves behind. */
mrb_bool mrb_str_valid_encoding_p(mrb_state *mrb, mrb_value str);

Comment only; no code changes.

Verified

• rake test on a full-core build: 2246 tests, all green
• prek run --all-files passes, except that markdownlint could not install
  locally (npm engine mismatch); no Markdown is touched here

Summary by CodeRabbit

• Documentation
  • Clarified documentation for string encoding validation, including behavior for binary strings and non-UTF-8 encodings.