[PERF] Don't spawn so many compilers (3/2) (19m -> 250k)

[blyxyas]

Optimize needless_doctest_main, make it short-circuit, make sure that we don't spin up a new compiler on EVERY code block.

---

The old implementation was creating a new compiler, new parser, new thread, new SessionGlobals, new everything for each code block. No matter if they actually didn't even contain fn main() or anything relevant.

On callgrind, seems that we're reducing about a 6.7242% de cycle count (which turns out to be a 38 million instruction difference, great!). Benchmarked in bumpalo-3.16.0. Also on bumpalo we spawn 78 less threads. This moves SessionGlobals::new from the top time-consuming function by itself in some benchmarks, into one not even in the top 500.

Also, populate the test files.

changelog:[needless_doctest_main]: Avoid spawning so many threads in unnecessary circumstances

[rustbot]

r? @y21

rustbot has assigned @y21.
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

[matthiaskrgr]

hm, if someone writes fn main() instead of fn main() will this cause false negatives?

[blyxyas]

Fixed (and improved filtering!) Now we check for instances of fn (two times) and main in general.

[y21]

That's a nice optimisation, but I have one question

[y21]

The splitn condition is a bit confusing, I can't tell what it's checking for. splitn(2) limits the iterator to 2 items, so isn't nth(2) always None?

Is this meant to be splitn(3, ..).nth(2).is_none() to check that there are at most two occurences of fn? But even then, I don't see why this is needed

[blyxyas]

Oops, what an oversight that splitn(2) should actually be splitn(3), I'll add tests with 3 and 4 functions.

This is necessary because in the actual check_code_sample we care only if the code block has one function and that function is main() (This was the behaviour even before this PR)

If there's more than one function, fn main would be relevant as a separate entity so we cannot report it as useless.

[blyxyas]

I've simplified filtering again with the risk of being over-scoped. This is still a pretty heavy optimization just with some more flexibility. Also accounting for all the possible spaces in between fn and main.