[stdlib] Remove additional customization points

[airspeedswift]

This removes some unnecessary customization points on standard library protocols:

map, filter and forEach on Sequence

It is unrealistic that a collection would be able to implement map (which takes a transformation closure) to create a mapped array faster than the default implementation.

While it is possible that a type's forEach implementation might be able to eek out a small performance benefit (for example, to avoid the reference count bump of putting self into an iterator), it is generally harmful to encourage this kind of "maybe forEach could be faster" micro-optimization (for example, see here where error control flow was used in order to break out of the for-each early).

first on Collection and last on BidirectionalCollection

It's important to remove these in order for collections of move-only types to be able to implement these protocols. Returning a single element of the collection inside an optional would otherwise need to consume the entire collection.

While it is conceivable that faster implementations of first for particular collections could exist, it's unlikely to be material and experience in benchmarking so far suggests that where they did exist, those implementations were actually slower.

Amended as brought up on the review thread to add more entries:

suffix(from:), prefix(upTo:) and prefix(through:)

These are trivial and really don't need to be customizable.

[airspeedswift]

@swift-ci please benchmark

[swift-ci]

Build comment file:

Performance: -O

TEST	OLD	NEW	DELTA	RATIO
Improvement
IterateData	1742	1621	-6.9%	1.07x (?)

Code size: -O

TEST	OLD	NEW	DELTA	RATIO
Improvement
NibbleSort.o	14536	12448	-14.4%	1.17x
COWTree.o	14900	14660	-1.6%	1.02x
LuhnAlgoLazy.o	18256	18016	-1.3%	1.01x
LuhnAlgoEager.o	18258	18018	-1.3%	1.01x
SequenceAlgos.o	23379	23139	-1.0%	1.01x

Performance: -Osize

TEST	OLD	NEW	DELTA	RATIO
Improvement
IterateData	1806	1591	-11.9%	1.14x
RemoveWhereFilterString	340	312	-8.2%	1.09x

Code size: -Osize

TEST	OLD	NEW	DELTA	RATIO
Regression
StringRemoveDupes.o	8973	9077	+1.2%	0.99x
Improvement
NibbleSort.o	19752	17000	-13.9%	1.16x
COWTree.o	14078	13838	-1.7%	1.02x
LuhnAlgoLazy.o	15920	15680	-1.5%	1.02x
LuhnAlgoEager.o	15922	15682	-1.5%	1.02x

Performance: -Onone

TEST	OLD	NEW	DELTA	RATIO
Improvement
StringInterpolationSmall	6571	6110	-7.0%	1.08x

Code size: Swift libraries

TEST	OLD	NEW	DELTA	RATIO
Improvement
libswiftCoreAudio.dylib	32768	28672	-12.5%	1.14x
libswiftsimd.dylib	286720	266240	-7.1%	1.08x
libswiftCloudKit.dylib	102400	98304	-4.0%	1.04x
libswiftCore.dylib	3969024	3858432	-2.8%	1.03x
libswiftFoundation.dylib	1839104	1810432	-1.6%	1.02x

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

---

[airspeedswift]

@swift-ci please test

[airspeedswift]

@swift-ci please test

[airspeedswift]

@swift-ci please smoke test

[airspeedswift]

@swift-ci please smoke test

[airspeedswift]

@swift-ci please smoke test

[airspeedswift]

@swift-ci please smoke test linux platform

[airspeedswift]

@swift-ci please smoke test linux platform

[airspeedswift]

@swift-ci please test linux platform

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - a211184de1473beca456847814d67f886784f5cf