[semantic-arc-opts] Teach load [copy] -> load_borrow about forwarding…

[gottesmm]

… instructions.

I implemented this some time ago for copy_value, but I (IIRC due to time) did
not enable it for the load [copy] optimization. Specifically now we should be
able to optimize this:

%ref = ref_element_addr %guaranteedArg
%0a = load [copy] %ref
%0b = unchecked_ref_cast %0a
destroy_value %0b

->

%ref = ref_element_addr %guaranteedArg
%0a = load_borrow %guaranteedArg
%0b = unchecked_ref_cast %0a
end_borrow %0a

[gottesmm]

@swift-ci smoke test

[gottesmm]

NOTE: There are two commits here, the first a refactor the 2nd the meat of the transform.

[gottesmm]

@swift-ci smoke test

[gottesmm]

@swift-ci smoke test

[gottesmm]

@swift-ci smoke test

[gottesmm]

@swift-ci benchmark

[gottesmm]

@swift-ci benchmark

[gottesmm]

@swift-ci benchmark

[gottesmm]

@swift-ci benchmark

[gottesmm]

NOTE: I split out the first commit into a separate PR since it is pretty trivial.

[gottesmm]

Here: #26958

[swift-ci]

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	4431	5732	+29.4%	0.77x (?)
Improvement	OLD	NEW	DELTA	RATIO
NIOChannelPipeline	169	85	-49.7%	1.99x
DropFirstArrayLazy	14	13	-7.1%	1.08x (?)

Code size: -O

Improvement	OLD	NEW	DELTA	RATIO
NIOChannelPipeline.o	4096	4048	-1.2%	1.01x

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
DataAppendDataLargeToLarge	26600	38000	+42.9%	0.70x (?)
Improvement	OLD	NEW	DELTA	RATIO
NIOChannelPipeline	174	91	-47.7%	1.91x
SuffixCountableRange	5	4	-20.0%	1.25x (?)

Code size: -Osize

Performance: -Onone

Code size: -swiftlibs

Improvement	OLD	NEW	DELTA	RATIO
libswiftCloudKit.dylib	73728	69632	-5.6%	1.06x

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

[gottesmm]

@weissi check out that NIOChannelPipeline improvement.

[weissi]

Wow, that is amazing!!! Thank you

[gottesmm]

@swift-ci smoke test and merge

[gottesmm]

@swift-ci smoke test and merge

[gottesmm]

@swift-ci smoke test and merge

[gottesmm]

@swift-ci smoke test and merge