Fix system param validation for piped systems

[alice-i-cecile]

Objective

• Piped systems are an edge case that we missed when reworking system parameter validation.
• Fixes Single causes piped systems to panic #18755.

Solution

• Validate the parameters for both systems, combining the errors if both failed validation by simply using an early out.
• Also fix the same bug for combinator systems while we're here.

Testing

I've added a large number of tests checking the behavior under various permutations. These are separate tests, rather than one mega test because a) it's easier to track down bugs that way and b) many of these are should_panic tests, which will halt the evaluation of the rest of the test!

I've also added a test for exclusive systems being pipeable because we don't have one and I was very surprised that that works!

[MiniaczQ]

Looks fine, no point spending more time on this, since the full fix requires checking params inside each System::run_unsafe, where the abstraction of chain will be gone.

[chescock]

Won't doing eager validation like this re-introduce the issues from #16638 (comment)? If a user pipes a system that inserts a resource into a system with Res, then that will succeed today but would fail under this change. (And if they pipe a system that invalidates a Single into one that uses it, then the second system will fail instead of skipping, but that's no worse than now.)

[alice-i-cecile]

Won't doing eager validation like this re-introduce the issues from #16638 (comment)? If a user pipes a system that inserts a resource into a system with Res, then that will succeed today but would fail under this change

I was going to say that this is quite unlikely, since all of our existing validated parameters requires structural changes to invalidate, but system piping doesn't work for exclusive systems. I wrote a test: turns out at some point that changed!

That said, I think this is a better tradeoff for now: piping exclusive systems or doing custom validated system params that are affected by an earlier system in the pipe feels quite rare. By contrast, the existing bug breaks a relatively common pattern (using validated params in a piped system) by changing the "everything is fine" behavior into a confusing panic.

[chescock]

Yeah, breaking resource_exists::<MyRes>.and(do_stuff) is what seemed really bad, but I see you're doing this for pipe but not combinators, so this does seem like the right tradeoff!

[alice-i-cecile]

Yeah, breaking resource_exists::.and(do_stuff) is what seemed really bad

Yeah, I saw that doc test fail, stared at it for a while in dismay, and then gave up and reverted the change. That is a much harder fix than I want for "last minute during 0.16".

[MrGVSV]

Not an ECS dev but the changes look good to me and has really solid tests.