gh-145122: Add _GUARD_NOS_ANY_DICT to prevent STORE_SUBSCR_DICT on frozendict

[corona10]

• Issue: Specialized STORE_SUBSCR_DICT allows silent mutation of frozendict after adaptive specialization #145122

[corona10]

cc @markshannon@Fidget-Spinner@vstinner

[corona10]

I am not sure this is correct way to handle this.
Should I just follow the same as _GUARD_TOS_ANY_SET which does not call sym_set_type?

[corona10]

Without this PR, tthe est code will fail like this:

Raised RLIMIT_NOFILE: 256 -> 1024
Using random seed: 2311893955
0:00:00 load avg: 7.54 Run 1 test sequentially in a single process
0:00:00 load avg: 7.54 [1/1] test_opcache
test test_opcache failed -- Traceback (most recent call last):
File "/Users/dongheena/oss/cpython/Lib/test/test_opcache.py", line 1910, in test_store_subscr
with self.assertRaises(TypeError):
~~~~~~~~~~~~~~~~~^^^^^^^^^^^
AssertionError: TypeError not raised
0:00:00 load avg: 7.54 [1/1/1] test_opcache failed (1 failure)

[vstinner]

This guard is no longer needed, it can be removed, no?

[corona10]

You’re right, I think we can remove the guard. However, let’s wait for Ken and Mark’s opinions on whether deleting it would be safe (I think yes), or if they prefer to leave it as is for potential future use.

[Fidget-Spinner]

If a uop is no longer needed it's best to remove it so that we don't leave things to bitrot in the code.

[corona10]

Thanks! I remove the _GUARD_TOS_DICT

[corona10]

@Fidget-Spinner Could you take a look? The fix in bytecode.c looks good to me, but I’d appreciate your review of the changes in optimized_bytecode.c.