[String] String-from-whole-Substring fast-path.

[milseman]

Add in a fast-path for Strings created from Substring which covers the
entire String. Put String-from-Substring behind a non-inlinable
resilience barrier for future flexibility.

[milseman]

@swift-ci please benchmark

[milseman]

@swift-ci please test

[swift-ci]

Build comment file:

Performance: -O

TEST	OLD	NEW	DELTA	RATIO
Regression
CharacterLiteralsLarge	97	108	+11.3%	0.90x
IterateData	1396	1509	+8.1%	0.93x
Improvement
StringFromLongWholeSubstring	80	11	-86.2%	7.27x
StringFromLongWholeSubstringGeneric	91	22	-75.8%	4.14x
Breadcrumbs.CopyUTF16CodeUnits.Mixed	68	55	-19.1%	1.24x
FlattenListLoop	4336	3970	-8.4%	1.09x
MapReduceAnyCollection	398	369	-7.3%	1.08x
FlattenListFlatMap	6864	6398	-6.8%	1.07x (?)
MapReduce	425	397	-6.6%	1.07x

Code size: -O

TEST	OLD	NEW	DELTA	RATIO
Improvement
StringEdits.o	13063	11967	-8.4%	1.09x
Substring.o	20263	19167	-5.4%	1.06x
CSVParsing.o	31641	30257	-4.4%	1.05x
WordCount.o	44211	43115	-2.5%	1.03x
StringComparison.o	44486	43390	-2.5%	1.03x

Performance: -Osize

TEST	OLD	NEW	DELTA	RATIO
Regression
IterateData	1353	1499	+10.8%	0.90x
Improvement
StringFromLongWholeSubstring	85	11	-87.1%	7.73x
StringFromLongWholeSubstringGeneric	90	23	-74.4%	3.91x
Breadcrumbs.CopyUTF16CodeUnits.Mixed	68	55	-19.1%	1.24x
InsertCharacterEndIndexNonASCII	59	55	-6.8%	1.07x (?)

Code size: -Osize

TEST	OLD	NEW	DELTA	RATIO
Improvement
CSVParsing.o	31977	30697	-4.0%	1.04x

Performance: -Onone

TEST	OLD	NEW	DELTA	RATIO
Improvement
StringFromLongWholeSubstring	81	12	-85.2%	6.75x
StringFromLongWholeSubstringGeneric	103	29	-71.8%	3.55x
ArrayOfPOD	860	775	-9.9%	1.11x (?)

Code size: -swiftlibs

TEST	OLD	NEW	DELTA	RATIO
Improvement
libswiftNetwork.dylib	163840	159744	-2.5%	1.03x

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 Pro
  Model Identifier: MacPro6,1
  Processor Name: 12-Core Intel Xeon E5
  Processor Speed: 2.7 GHz
  Number of Processors: 1
  Total Number of Cores: 12
  L2 Cache (per Core): 256 KB
  L3 Cache: 30 MB
  Memory: 64 GB

--------------