CopyPropagation: Avoid regenerating destroys.

[atrick]

CanonicalizeOSSA is now used iteratively in SILCombine. To avoid
endless worklist iteration based on whether InstructionDeleter's
callbacks fired, ensure that destroys are only deleted and recreated
when necessary.

[atrick]

@swift-ci test

[atrick]

@swift-ci benchmark

[atrick]

@swift-ci smoke test

[swift-ci]

Performance (x86_64): -O

Regression	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	2397	2653	+10.7%	0.90x (?)
Improvement	OLD	NEW	DELTA	RATIO
DataSubscriptSmall	23	15	-34.8%	1.53x
Data.append.Sequence.64kB.Count.I	44	29	-34.1%	1.52x
Data.append.Sequence.64kB.Count	44	29	-34.1%	1.52x
DataCreateSmallArray	2250	1600	-28.9%	1.41x
Data.append.Sequence.809B.Count.I	80	62	-22.5%	1.29x
Data.append.Sequence.809B.Count	79	62	-21.5%	1.27x (?)
Data.init.Sequence.64kB.Count.RE.I	20	16	-20.0%	1.25x
Data.init.Sequence.64kB.Count.RE	20	16	-20.0%	1.25x (?)
Data.init.Sequence.809B.Count.RE.I	43	38	-11.6%	1.13x (?)
Data.init.Sequence.809B.Count.RE	42	38	-9.5%	1.11x (?)
Data.init.Sequence.64kB.Count0.RE.I	184	168	-8.7%	1.10x (?)
Data.init.Sequence.64kB.Count0.RE	183	168	-8.2%	1.09x (?)
StringToDataSmall	750	700	-6.7%	1.07x (?)

Code size: -O

Improvement	OLD	NEW	DELTA	RATIO
DataBenchmarks.o	61878	59931	-3.1%	1.03x
UTF8Decode.o	24019	23381	-2.7%	1.03x
ObjectiveCBridging.o	59073	58209	-1.5%	1.01x

Performance (x86_64): -Osize

Regression	OLD	NEW	DELTA	RATIO
Data.append.Sequence.64kB.Count.RE.I	20	29	+45.0%	0.69x
Data.append.Sequence.64kB.Count.RE	20	29	+45.0%	0.69x
DataAppendSequence	5500	7000	+27.3%	0.79x (?)
ObjectiveCBridgeStubDateAccess	130	152	+16.9%	0.86x (?)
DataReplaceMediumBuffer	2600	2900	+11.5%	0.90x (?)
DataCreateEmptyArray	900	1000	+11.1%	0.90x (?)
ProtocolDispatch	239	261	+9.2%	0.92x (?)
Improvement	OLD	NEW	DELTA	RATIO
Data.append.Sequence.64kB.Count.I	58	30	-48.3%	1.93x
Data.append.Sequence.64kB.Count	58	30	-48.3%	1.93x
Data.init.Sequence.64kB.Count	58	30	-48.3%	1.93x
Data.init.Sequence.2047B.Count.I	98	52	-46.9%	1.88x
Data.init.Sequence.2049B.Count.I	98	52	-46.9%	1.88x
Data.init.Sequence.64kB.Count.I	58	32	-44.8%	1.81x
Data.init.Sequence.809B.Count	84	50	-40.5%	1.68x
Data.init.Sequence.809B.Count.I	84	50	-40.5%	1.68x
Data.init.Sequence.511B.Count.I	92	60	-34.8%	1.53x
Data.init.Sequence.64kB.Count.RE.I	29	19	-34.5%	1.53x
Data.init.Sequence.64kB.Count.RE	29	19	-34.5%	1.53x
Data.init.Sequence.513B.Count.I	93	61	-34.4%	1.52x
Data.append.Sequence.809B.Count	98	65	-33.7%	1.51x
Data.append.Sequence.809B.Count.I	98	65	-33.7%	1.51x
DataSubscriptSmall	26	19	-26.9%	1.37x
Data.init.Sequence.809B.Count.RE	55	42	-23.6%	1.31x
Data.init.Sequence.809B.Count.RE.I	55	42	-23.6%	1.31x
DataCountMedium	17	15	-11.8%	1.13x (?)

Code size: -Osize

Improvement	OLD	NEW	DELTA	RATIO
DataBenchmarks.o	52105	50934	-2.2%	1.02x

Performance (x86_64): -Onone

Improvement	OLD	NEW	DELTA	RATIO
DataReplaceLargeBuffer	12	11	-8.3%	1.09x (?)

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