noop in setTypeAcquisiton for undefined

[jessetrinity]

#40952 mistakenly sets typeAcquisition to {} if it isn't explicitly set by the editor, disabling type acquisition in inferred projects. This reverts the changes made to removeLocalTypingsFromTypeAcquisition in that PR and noops in setTypeAcquisition when newTypeAcquisition is undefined.

Issue was caught in VS nightly integration tests.

[sheetalkamat]

The change looks correct but i am surprised we dont have test. can you add one please so we dont regress this in future. Thanks

[minestarks]

Looks good. +1 to adding test if possible

[minestarks]

@DanielRosenwasser heads up - we want this for 4.1 RC, looks like it's almost ready to go. Thanks!

[jessetrinity]

@sheetalkamat I've been looking at the tests for a while and I can't figure out what we should actually have as a regression test. Should there be a unit test for the compilerOptions/watchOptions/typeAcquisition setters to make sure they noop on undefined?

[jessetrinity]

To clarify - inferred projects infer typeAcquisition when it is undefined, so my previous change caused the inference to never happen (the getter started returning {}). A minimal test would basically be:

project.setTypeAcquisition(undefined);
assert.isTrue(project.getTypeAcquisition().enable, true)

[sheetalkamat]

assert.deepEqual(project.getTypeAcquisition(), expected) is what i think better test.