Prune the canonical lint set

[DJMcNab]

We've had complaints that the lint set is too strict, which I think is fair. As such, I suggest that we remove some of the most pedantic lints. It also adds a new set which we "want to" follow, but which don't block PRs.
It's expected that we would work to not regress these between (for example) crate releases.

Of note:

1. allow_attributes is removed from the default set. This doesn't mean we shouldn't still aim to use expect, but having an allow is fine.
2. default_trait_access is newly added to the new "periodic" category.
3. missing_panics_doc is removed entirely.
4. wildcard_imports is removed entirely

[PoignardAzur]

LGTM.

[xStrom]

Looks good.

I think removing match_same_arms from the default set is particularly good. It's a good lint some times, but we also have existing code where it clearly makes things worse.

[xStrom]

On further thought, are there examples of issues with clippy::doc_markdown? I haven't looked deeply into it, but from memory it has been a good experience.

@waywardmonkeys you've been the biggest implementor of this. What's your take, should we keep it on by default?

[waywardmonkeys]

I think clippy::doc_markdown should stay on by default.

[waywardmonkeys]

We can also have CI run a job periodically (every Monday morning?) that runs the periodic checks.

[DJMcNab]

I removed doc_markdown as:

1. it strikes me as the sort of thing which someone running into it in CI would find extremely annoying
2. It's really simple to do drive-by fixes for (hence being in the editor category)

Certainly happy to optimistically restore it, though.

A weekly check with a Zulip webhook message for the periodic checks might be nice. I suspect that's too much effort for the amount of gain (i.e. that re-adding them to CI proper would probably be better than that...), but I could be convinced.