[SkipFunctionBodies] Build everything but the stdlib with function body skipping

[harlanhaskins]

Per a conversation with @atrick, see what the impact is of building everything but the stdlib with function body skipping, so we can see if the issues are stdlib-specific. Granted, not many of the benchmarks use the overlays...

[harlanhaskins]

@swift-ci please benchmark

[swift-ci]

Build failed before running benchmark.

[harlanhaskins]

@swift-ci please benchmark

[swift-ci]

Performance: -O

Improvement	OLD	NEW	DELTA	RATIO
ObjectiveCBridgeStubNSDataAppend	1660	1520	-8.4%	1.09x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
DropWhileSequence	13	14	+7.7%	0.93x (?)
Improvement	OLD	NEW	DELTA	RATIO
ArrayAppendGenericStructs	1520	590	-61.2%	2.58x (?)
ObjectiveCBridgeStubFromNSStringRef	91	85	-6.6%	1.07x (?)

Code size: -Osize

Performance: -Onone

Improvement	OLD	NEW	DELTA	RATIO
ArrayAppendGenericStructs	650	600	-7.7%	1.08x (?)

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

[harlanhaskins]

@swift-ci please test

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - 5301fa0

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - 5301fa0

[atrick]

No problems with the benchmarks. The standard library needs to be built specially anyway. For example, we can't parallelize the LLVM passes as we usually do because we too much benefit from cross-file optimization. I think it's the same with serialization. I don't expect regular frameworks to be so sensitive to this issue. Now, it's interesting that with resilience the compiler isn't supposed to be making any assumptions about the non-inlinable code, so it's probably worth following up on the compiler side to understand why it does things differently. But there's no harm done in this PR.