Fix generic_ivar_set_shape_ivptr for table rebuild

[Bug #21438]

Previously GC could trigger a table rebuild of the generic ivar
st_table in the middle of calling the st_update callback. This could
cause entries to be reallocated or rearranged and the update to be for
the wrong entry.

This commit adds an assertion to make that case easier to detect, and
replaces the st_update with a separate st_lookup and st_insert.

Also free after insert in generic_ivar_set_shape_ivptr

Previously we were performing a realloc and then inserting the new value
into the table. If the table was flagged as requiring a rebuild, this
could trigger GC work and marking within that GC could access the ivptr
freed by realloc.

Co-authored-by: John Hawthorn <john@hawthorn.email>
Co-authored-by: Aaron Patterson <tenderlove@ruby-lang.org>

[Backport #21094]

The keys are IDs, so we cannot update references on them.

So that it doesn't get included in the generated binaries for builds
that don't support loading shared GC modules

Co-Authored-By: Peter Zhu <peter@peterzhu.ca>

    ASAN_OPTIONS="detect_leaks=1" RUBY_FREE_AT_EXIT=1 ./miniruby -e 'trace_var(:$x){}'

* Use FL_USER0 for ELTS_SHARED

This makes space in RString for two bits for chilled strings.

* Mark strings returned by `Symbol#to_s` as chilled

[Feature #20350]

`STR_CHILLED` now spans on two user flags. If one bit is set it
marks a chilled string literal, if it's the other it marks a
`Symbol#to_s` chilled string.

Since it's not possible, and doesn't make much sense to include
debug info when `--debug-frozen-string-literal` is set, we can't
include allocation source, but we can safely include the symbol
name in the warning message, making it much easier to find the source
of the issue.

Co-Authored-By: Étienne Barrié <etienne.barrie@gmail.com>

---------

Co-authored-by: Étienne Barrié <etienne.barrie@gmail.com>
Co-authored-by: Jean Boussier <jean.boussier@gmail.com>