[opt] promote stack alloc_ref to object

[zoecarver]

Based on #30736. This patch updates StackPromotion to promote alloc_refs on the stack to object instructions. Given that there's currently no support for object instructions in the optimizer, this pass shouldn't have many (if any) performance improvements. It just lays the groundwork for future performance improvements.

In the future, the promotion could probably be more aggressive but, as of now, it will only promote stack alloc_refs so there shouldn't be any regression.

[zoecarver]

@swift-ci please benchmark

[zoecarver]

As said in the summary, I don't expect any performance improvements. That's just to check for regressions (which shouldn't be possible because stack alloc_refs and objects both allocate objects in the same way).

[swift-ci]

Build failed before running benchmark.

[zoecarver]

@swift-ci please benchmark

[swift-ci]

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
Hanoi	2110	2430	+15.2%	0.87x
ObjectAllocation	123	135	+9.8%	0.91x (?)
Improvement	OLD	NEW	DELTA	RATIO
ObserverPartiallyAppliedMethod	1360	820	-39.7%	1.66x
ObserverClosure	1350	820	-39.3%	1.65x

Code size: -O

Improvement	OLD	NEW	DELTA	RATIO
Hanoi.o	3120	3056	-2.1%	1.02x
SortArrayInClass.o	2738	2690	-1.8%	1.02x
ObserverForwarderStruct.o	2852	2804	-1.7%	1.02x
RecursiveOwnedParameter.o	2270	2238	-1.4%	1.01x
DictionaryBridge.o	3420	3372	-1.4%	1.01x
ObserverUnappliedMethod.o	5089	5025	-1.3%	1.01x

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
ObjectiveCBridgeStubToNSDateRef	2120	2320	+9.4%	0.91x (?)
ObjectiveCBridgeStubToNSDate2	330	360	+9.1%	0.92x (?)
ObjectAllocation	135	147	+8.9%	0.92x (?)
Improvement	OLD	NEW	DELTA	RATIO
FlattenListLoop	3263	2790	-14.5%	1.17x (?)
String.data.LargeUnicode	71	65	-8.5%	1.09x (?)

Code size: -Osize

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
DataCountSmall	47	51	+8.5%	0.92x (?)
DataSetCountSmall	84	91	+8.3%	0.92x (?)

Code size: -swiftlibs

How to read the data

The tables contain differences in performance which are larger than 8% and differences in code size which are larger than 1%.

If you see any unexpected regressions, you should consider fixing the
regressions before you merge the PR.

Noise: Sometimes the performance results (not code size!) contain false
alarms. Unexpected regressions which are marked with '(?)' are probably noise.
If you see regressions which you cannot explain you can try to run the
benchmarks again. If regressions still show up, please consult with the
performance team (@eeckstein).

Hardware Overview

  Model Name: Mac mini
  Model Identifier: Macmini8,1
  Processor Name: 6-Core Intel Core i7
  Processor Speed: 3.2 GHz
  Number of Processors: 1
  Total Number of Cores: 6
  L2 Cache (per Core): 256 KB
  L3 Cache: 12 MB
  Memory: 64 GB

[eeckstein]

See my comment in #30736

[zoecarver]

I'm closing this given the discussion in #30787 and because #30810 had significantly better performance improvements.