This changes `MCRegUnit` type from `unsigned` to `enum class : unsigned`
and inserts necessary casts.
The added `MCRegUnitToIndex` functor is used with `SparseSet`,
`SparseMultiSet` and `IndexedMap` in a few places.

`MCRegUnit` is opaque to users, so it didn't seem worth making it a
full-fledged class like `Register`.

Static type checking has detected one issue in
`PrologueEpilogueInserter.cpp`, where `BitVector` created for
`MCRegister` is indexed by both `MCRegister` and `MCRegUnit`.

The number of casts could be reduced by using `IndexedMap` in more
places and/or adding a `BitVector` adaptor, but the number of casts *per
file* is still small and `IndexedMap` has limitations, so it didn't seem
worth the effort.

Pull Request: https://github.com/llvm/llvm-project/pull/167943

(original PR: #166210)

This allows more accurate alias analysis to apply at the bundle level.
This has a bunch of minor effects in post-RA scheduling that look mostly
beneficial to me, all of them in AMDGPU (the Thumb2 change is cosmetic).

The pre-existing (and unchanged) test in
CodeGen/MIR/AMDGPU/custom-pseudo-source-values.ll tests that MIR with a
bundle with MMOs can be parsed successfully.

v2:
- use cloneMergedMemRefs
- add another test to explicitly check the MMO bundling behavior

v3:
- use poison instead of undef to initialize the global variable in the
test

v4:
- treat bundle memory accesses as never trivially disjoint

The debug info attached to the BUNDLE is the first instruction in the
BUNDLE, even if a better debug info (line:column) is present in the
later instructions of the bundle. The patch tries to get a better debug
info first. If not, then a worse debug info without line number is
chosen.

---------

Co-authored-by: Vladislav Dzhidzhoev <dzhidzhoev@gmail.com>
Co-authored-by: Orlando Cazalet-Hyams <orlandoch.och@gmail.com>

Reverts llvm/llvm-project#166210

Buildbot failures in the libc on GPU bot:
https://lab.llvm.org/buildbot/#/builders/10/builds/16711

This allows more accurate alias analysis to apply at the bundle level.
This has a bunch of minor effects in post-RA scheduling that look mostly
beneficial to me, all of them in AMDGPU (the Thumb2 change is cosmetic).

The pre-existing (and unchanged) test in
CodeGen/MIR/AMDGPU/custom-pseudo-source-values.ll tests that MIR with a
bundle with MMOs can be parsed successfully.

v2:
- use cloneMergedMemRefs
- add another test to explicitly check the MMO bundling behavior

v3:
- use poison instead of undef to initialize the global variable in the
test

This is in preparation of a future AMDGPU change where we are going to
create bundles before register allocation and want to rely on the
TwoAddressInstructionPass handling those bundles correctly.

v2:
- simplify the virtual register check and the test

A dead redefinition should override any earlier non-dead definition
inside a bundle.

Also remove KilledDefSet since it can be folded into DeadDefSet.

When tracking defs in finalizeBundle two sets are used. LocalDefs is
used to track defined virtual and physical registers, while LocalDefsP
is used to track defined register units for the physical registers.

This patch moves the updates of LocalDefsP to only iterate over regunits
when a new physical register is added to LocalDefs. When the physical
register already is present in LocalDefs, then the corresponding
register units are present in LocalDefsP. So it was a waste of time to
add them to the set again.

Cause:
1. `implicit_def` inside bundle does not count for define of reg in
machineinst verifier
2. Including `implicit_def` will cause relative reg not define, result
in `Bad machine code: Using an undefined physical register` in the
machineinst verifier

Fixes https://github.com/llvm/llvm-project/issues/139102

---------

Co-authored-by: Matt Arsenault <Matthew.Arsenault@amd.com>

Replace its only use in the AMDGPU R600 backend with a call to
finalizeBundles.