Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@crates/vm/src/builtins/str.rs` around lines 752 - 770, The new ICU-backed
casefold implementation (function casefold matching on PyKindStr::Ascii,
PyKindStr::Utf8, PyKindStr::Wtf8 and using CaseMapper/Wtf8Buf) lacks regression
tests; add focused unit tests that call the public string casefold behavior to
cover: an ASCII-only input (e.g., "ABC" -> "abc"), an expansion fold (e.g., "ß"
-> "ss"), Greek sigma cases (final vs medial sigma behavior such as "ΣςΣ" or
examples that exercise final sigma folding), and a WTF-8/lone-surrogate
preservation case (construct a Wtf8-containing string with a lone surrogate and
assert it is preserved after casefold). Put these tests in the existing str
tests module so they exercise all match arms (Ascii, Utf8 via CaseMapper, and
Wtf8 via Wtf8Buf) and assert exact output bytes/characters to catch regressions.