Consider adding a fix for `pytest-raises-ambiguous-pattern` (`RUF043`)

[ntBre]

It's certainly a more noisy rule and it's unfortunate we can't provide an auto fix (because Ruff can't know what the desired behavior is). But I think it's fine to stabilize

Unless we want to add a heuristic for the auto fix. E.g.:

• A string ending with . is probably a sentence -> escape it
• A string with a sequence of metacharacters (e.g. .* or .+) is probably a regex -> make it raw. We can limit it to only support some specific patterns.

The first fix would have to be unsafe because it does change the program's semantics. The second one could be safe but I think it's still best to make user explicitly opt-in. We could document the heuristics in the rule and having a fix could reduce the churn for users.

I'd postpone the promotion to stable if the suggested fix does sound reasonable (CC: @AlexWaygood)

Originally posted by @MichaReiser in #16657 (review)

[AlexWaygood]

We discussed when implementing the rule originally whether it should have an autofix or not and concluded "no" because adding backslash escapes may often be unwelcome (users will often prefer to make it a raw string), but it would probably be hard for Ruff to determine whether a raw string or an escape would be preferred. #14966 (comment)

I also don't love the idea of adding so many heuristics to an autofix; I think it'll be harder for users to understand without reading the docs (and I doubt people go and read the docs until a rule goes and does something that they don't expect).

So I'd be inclined to leave the rule as it is and stabilise it. That said, I don't have a strong opinion; happy to defer to you and @MichaReiser if you think users would find an unsafe fix helpful here!

[MichaReiser]

I don't feel strongly either and I agree that it can be confusing why the rule only suggests some fixes. However, considering how common the sentence ending with a dot is, providing a fix at least for that still seems useful to me.

(users will often prefer to make it a raw string), but it would probably be hard for Ruff to determine whether a raw string or an escape would be preferred. #14966 (comment)

I would keep it simple and always suggest re.escape because it works for all characters and the entire string was probably not meant to be a regex. However, I can see how this comes down to personal preferences.