IRGen: Use known value witnesses in generic metadata patterns.

[jckarter]

We were unnecessarily conservative here; generic metadata patterns support indirectable references
to value witness tables exported by the standard library, so if we have a fixed-layout generic
type that matches a known value witness table layout, use that instead of generating a new
value witness table.

[jckarter]

@swift-ci Please benchmark

[jckarter]

@swift-ci Please test

[jckarter]

@swift-ci Please test Windows

[swift-ci]

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	3938	4567	+16.0%	0.86x (?)
Set.isDisjoint.Seq.Empty.Box	78	89	+14.1%	0.88x (?)
Set.subtracting.Empty.Box	8	9	+12.5%	0.89x (?)
Set.isDisjoint.Seq.Box.Empty	68	75	+10.3%	0.91x
RemoveWhereFilterInts	23	25	+8.7%	0.92x (?)
SuffixSequence	289	314	+8.7%	0.92x (?)
SuffixSequenceLazy	289	314	+8.7%	0.92x
UTF8Decode_InitDecoding	105	114	+8.6%	0.92x (?)
Set.isStrictSubset.Seq.Int.Empty	119	129	+8.4%	0.92x (?)
Set.isSubset.Seq.Int.Empty	118	127	+7.6%	0.93x (?)
Improvement	OLD	NEW	DELTA	RATIO
CharIndexing_punctuated_unicodeScalars_Backwards	800	720	-10.0%	1.11x (?)
NormalizedIterator_fastPrenormal	640	590	-7.8%	1.08x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
FlattenListLoop	2168	2748	+26.8%	0.79x (?)
Set.isStrictSuperset.Seq.Empty.Int	169	183	+8.3%	0.92x (?)
Improvement	OLD	NEW	DELTA	RATIO
RemoveWhereMoveInts	21	18	-14.3%	1.17x
DistinctClassFieldAccesses	195	180	-7.7%	1.08x (?)
NSStringConversion.Long	525	490	-6.7%	1.07x (?)

Code size: -Osize

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
ArrayOfGenericPOD2	713	834	+17.0%	0.85x (?)
UTF8Decode_InitDecoding	126	138	+9.5%	0.91x (?)

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: 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