Merge rules checking the same violation

[not-my-profile]

As per the rule naming convention established in #2712 the rule name should be the bad thing that's being checked for and not contain the solution, so we should merge the following rules accordingly:

• Merge use-contextlib-suppress (SIM105) and try-except-pass (S110) to try-except-pass
• Merge blind-except (BLE001) with bare-except (E722)
• Merge if-expr-with-true-false (SIM210) and if-expr-with-false-true (SIM211) into needless-bool (SIM103)
• Merge EM101, EM102 and EM103
• Merge convert-loop-to-any & convert-loop-to-all to reimplemented-builtin #2903
• Apparent duplicate rule (S307 and PGH001) #7283
• Rule duplication F509 and PLE1300 #11403

In the long run it would be nice if ruff supported different autofixes for one violation, see #2713.

[charliermarsh]

Agreed. Not sure which to "remove" (i.e., treat as the canonical rule, as opposed to the redirect), but I guess that gets back to the discussion in #2517.

[not-my-profile]

I don't think it matters much which code we map and which one we redirect for now, since #2517 should land soon(tm).

[charliermarsh]

To unify blind- and bare-except, we need to change bare-except to flag except Exception and except BaseException. I assume that's fine.

[not-my-profile]

Ah right ... good catch! I think except: and except BaseException: are equivalent since every exception must derive BaseException. except Exception is different though since some exceptions e.g. GeneratorExit, KeyboardInterrupt and SystemExit don't derive from it. So we may want to split blind-except into except-any and except-exception.

[charliermarsh]

We could also make it a setting on the rule, rather than creating separate rules. But I don’t think we’ve articulated a strict philosophy either way there.

[andersk]

use-contextlib-suppress (SIM105) is intentionally much weaker than try-except-pass (S110), and it’s not unreasonable to enable the former without the latter. See #1948.