Implement init?<T:BinaryInteger>(exactly:T) for the stdlib FP types

[stephentyrone]

Previously these always went through the FloatingPoint-provided default implementation, which is not particularly efficient. Also try removing inlinable from the generic _convert hooks, since we probably never want to actually inline them (rdar://problem/64544503).

Follow-up on #32617

[stephentyrone]

Not able to test this locally at present, so we have to cross our fingers.

[stephentyrone]

@swift-ci please test

[stephentyrone]

@swift-ci please benchmark

[swift-ci]

Performance: -O

Regression	OLD	NEW	DELTA	RATIO
ObjectiveCBridgeStubToNSDate2	540	610	+13.0%	0.89x (?)
ObjectiveCBridgeStubToNSDateRef	3460	3900	+12.7%	0.89x (?)
StringComparison_nonBMPSlowestPrenormal	1540	1680	+9.1%	0.92x
Array2D	6688	7216	+7.9%	0.93x (?)
Improvement	OLD	NEW	DELTA	RATIO
ParseFloat.Float.Exp	14	13	-7.1%	1.08x (?)
RandomTree.insert.Unmanaged.fast	211	196	-7.1%	1.08x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
ObjectiveCBridgeStubToNSDateRef	3500	4020	+14.9%	0.87x (?)
UTF8Decode_InitFromData_ascii_as_ascii	628	712	+13.4%	0.88x (?)
Array2D	6928	7520	+8.5%	0.92x
StringComparison_nonBMPSlowestPrenormal	1560	1690	+8.3%	0.92x
StringComparison_emoji	836	900	+7.7%	0.93x (?)
Improvement	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	6820	5973	-12.4%	1.14x (?)
DataAccessBytesMedium	102	94	-7.8%	1.09x (?)
RandomTree.insert.Unmanaged.fast	211	197	-6.6%	1.07x (?)

Code size: -Osize

Performance: -Onone

Regression	OLD	NEW	DELTA	RATIO
FrequenciesUsingReduceInto	4010	4340	+8.2%	0.92x (?)
ArrayOfPOD	1031	1113	+8.0%	0.93x (?)
Improvement	OLD	NEW	DELTA	RATIO
ObjectiveCBridgeStubFromNSDate	6840	5950	-13.0%	1.15x (?)

Code size: -swiftlibs

Improvement	OLD	NEW	DELTA	RATIO
libswiftCoreGraphics.dylib	65536	57344	-12.5%	1.14x

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