JIT: ensure fgFirstBB has appropriate flags

[AndyAyersMS]

This fixes an issue where release jits might sometimes generate bad GC info.
Keeping it minimal for now so we can consider servicing preview 8.

See #39023 for details.

[AndyAyersMS]

@BruceForstall PTAL
cc @dotnet/jit-contrib

Don't expect codegen diffs, will run SPMI in rel/rel mode to see if it spots any GC info diffs.

[AndyAyersMS]

Arm build failure looks spurious:

2020-07-28T19:57:03.4660344Z Building Custom Rule F:/workspace/_work/1/s/src/coreclr/src/inc/CMakeLists.txt
2020-07-28T19:57:03.4664353Z CMake does not need to re-run because F:/workspace/_work/1/s/artifacts/obj/coreclr/Windows_NT.arm.Checked/src/inc/CMakeFiles/generate.stamp is up-to-date.
2020-07-28T19:57:03.4988566Z MSBUILD : error MSB4017: The build stopped unexpectedly because of an unexpected logger failure.
2020-07-28T19:57:03.4990443Z Microsoft.Build.Exceptions.InternalLoggerException: The build stopped unexpectedly because of an unexpected logger failure. ---> System.ArgumentNullException: Parameter "projectProperty" cannot be null.
2020-07-28T19:57:03.4991907Z at Microsoft.Build.Shared.ErrorUtilities.VerifyThrowArgumentNull(Object parameter, String parameterName, String resourceName)
2020-07-28T19:57:03.4993161Z at Microsoft.Build.Collections.CopyOnWritePropertyDictionary`1.Set(T projectProperty)

[AndyAyersMS]

No release-release SPMI diffs on tests.pmi.Windows_NT.x64.Release.mch, or on the sample code I had from #39023. Not surprising -- it is unclear how often / how many methods might be affected by this, and SPMI might alter the odds.

[AndyAyersMS]

Format patch file is empty, no changes when I reformat locally.

[BruceForstall]

formatting: #40034

[AndyAyersMS]

Pretty sure all these failures are unrelated.