Added identity_assign_op lint

[JaafarTanoukhi]

changelog: [identity_assign_op]: new lint

fixes #16977

[rustbot]

r? @llogiq

rustbot has assigned @llogiq.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: 7 candidates
• 7 candidates expanded to 7 candidates
• Random selection from Jarcho, llogiq, samueltardieu

[samueltardieu]

There are a few problems with this proposed implementation:

• Ignoring the statement means removing it, not just replacing it with the LHS expression.
• Why is there a TODO in the source?
• User types may implement those operators, and they might not be removable because they can contain side effects.
• Floating points types are ignored.
• Even if the snippet cannot be retrieved, the lint is currently marked MachineApplicable which is not true.
• Tests involving macros should be added.

@rustbot author

View changes since this review

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[oscargus]

In the original issue there was a suggestion to focus on that this may be a bug, rather than that it can be simplified.

[rustbot]

This PR was rebased onto a different master commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

[JaafarTanoukhi]

@rustbot ready

[JaafarTanoukhi]

I’ve also updated this PR so that identity_assign_op does not lint identity assignments that are part of a series of otherwise similar assignment operations, as discussed in #13813.

If this approach looks good, I can follow up with a separate PR applying the same logic to identity_op.