DI: Correctly handle conditional destroy of 'self' in root class

[slavapestov]

In a designated initializer of a non-root class, 'self' becomes
fully initialized after the 'super.init' call, at which point
escaping uses of 'self' become valid, and releases of 'self' are
lowered to a 'strong_release' instruction, which runs the
deinitializer.

In a root class, 'self' becomes fully initialized after all stored
properties have been initialized, at which point escaping uses of
'self' become valid.

However, DI would still lower a conditional destroy of 'self' by
destroying any initialized stored properties and freeing the object
with 'dealloc_partial_ref'. This is incorrect, because 'self' may
have escaped.

In the non-conditional destroy case, we correctly lowered the
release to a 'strong_release' if all stored properties are known
to be initialized.

Fix DI to handle the conditional destroy case by first checking if all
bits in the control variable are set, and releasing the instance with
'strong_release' if so. The 'dealloc_partial_ref' is only emitted
if not at least one stored property was not initialized.

This ensures that we don't deallocate an instance that may have
escaped.

Fixes https://bugs.swift.org/browse/SR-13439, rdar://problem/67746791,
https://bugs.swift.org/browse/SR-13355, rdar://problem/67361228.

[slavapestov]

@jckarter Do you mind reviewing this? It's feels like a bit of a hack but I couldn't think of any other way of doing it. With a non-root class, we have a bit that records if super.init() was called to determine if the instance is fully initialized or not. With a root class however, the corresponding state transition happens when all stored properties become fully initialized, which can happen in multiple places in the control flow graph; it is also a dynamic property because of partial initialization.

We already did the right thing in the non-conditionally-initialized case by treating "all properties are initialized" as "object is fully initialized"; this just extends that to the dynamic case by checking if all bits in the control word are set. I did have to tweak the logic a little so that we now record a bit in the control word even for trivial fields; previously we could skip these since they did not affect the outcome, but now they do because we care about the "semantic" initialization state of the object.

[slavapestov]

Also, this changes runtime behavior of existing code, because the deinitializer will now run in more cases. I believe this is OK though, because the old behavior was untenable -- the deinitializer would run if DI could statically determine that all properties were initialized at the point of the release, otherwise it would never run. This is why the test case in the second JIRA failed, because the exit block had two incoming edges, one fully initialized and one partially initialized. It seemed silly that commenting out an if false { ... } block would change whether the deinitializer runs or not.

[slavapestov]

@swift-ci Please smoke test

[slavapestov]

@swift-ci Please test source compatibility