MC/DC Coverage re-introduction (1/3): THIR data gathering

[RenjiSann]

Tracking issue: rust-lang/goals#638

Identify MC/DC decision in the AST form, to allow for incoming instrumentation and mappings creation.

1. Introduce a -Z coverage-options=mcdc flag
2. Detect boolean expressions in THIR to:
   1. Gather sufficient data to construct MC/DC decisions;
   2. Emit block markers in MIR to track the expressions and instrument them

r? @davidtwco

[rustbot]

Some changes occurred in match lowering

cc @Nadrieril

Some changes occurred in coverage instrumentation.

cc @Zalathar

Some changes occurred in coverage instrumentation.

cc @Zalathar

[jyn514]

I'm quite confused that you've asked David to review this, who's not a maintainer of the coverage implementation, rather than Zalathar, who is.

[matthiaskrgr]

@jyn514 probably because David is the compiler champion for this goal https://rust-lang.github.io/goals/2026/mcdc-coverage-support.html

[Zalathar]

This PR series (?) seems to rely heavily on re-adding chunks of the old implementation that was specifically removed for imposing a heavy maintenance burden. I would not want that point to go unremarked.

The approach of doing any kind of non-trivial analysis during MIR building also seems like a dead end. It's already not working well for branch coverage, and MC/DC necessarily imposes a much higher complexity burden than branch coverage does.

[rust-bors]

☔ The latest upstream changes (presumably #162009) made this pull request unmergeable. Please resolve the merge conflicts by rebasing.