Fix idempotency issue when wrapping the last match arm in a block

[ytmimi]

Fixes #5193

The issues was related to the interaction between two differnt
configuration options when applied to the last match arm.

• trailing_comma = Never
• match_block_trailing_comma = true

Previously, rustfmt only consider the match_block_trailing_comma
configuration value when wrapping match arms. This lead to rustfmt
adding a trailing comma to all newly wrapped match arms.

Now, rustfmt also considers the trailing_comma configuration value
when deciding whether or not to add a trailing comma when wrapping the
last match arm in a block.

[ytmimi]

@calebcartwright Two things to note:

1. I plan on adding some additional tests cases to cover the various combinations of trailing_comma and match_block_trailing_comma. Just wanted to get the PR in when I identified an initial solution.

2. I know you mentioned that #5193 might be a duplicate. If you happen to track down the original issue (or other duplicates), I'll see if this PR fixes them as well and add test cases accordingly!

[calebcartwright]

Changes LGTM @ytmimi, and happy to merge as-is. Will hold off on merging since it sounds like there may be more you want to do, but lmk when you're ready

[ytmimi]

@calebcartwright Awesome! Yeah, I'd just want to add a few more test cases and see if I can track down any duplicates. I'll make sure to let you know when that's all taken care of!

[ytmimi]

I find it a little odd that trailing_comma=Never takes precedence over match_block_trailing_comma=true, for the last match arm but that's what I needed to do in order to prevent the idempotency issue since that's how arm_comma is implemented.

rustfmt/src/matches.rs

Lines 146 to 160 in a67d909

	fn arm_comma(config: &Config, body: &ast::Expr, is_last: bool) -> &'static str {
	if is_last && config.trailing_comma() == SeparatorTactic::Never {
	""
	} else if config.match_block_trailing_comma() {
	","
	} else if let ast::ExprKind::Block(ref block, _) = body.kind {
	if let ast::BlockCheckMode::Default = block.rules {
	""
	} else {
	","
	}
	} else {
	","
	}
	}

[ytmimi]

@calebcartwright no rush on this. I know it's been a while since you've last reviewed it. The last time you looked at these changes you said it was ready to merge, but since I've added some additional test cases I feel like it would be worth it to do another review if we choose to move forward with this