Prevent extra re-lowerings via "actual targets"

[seldridge]

Add a member to DependencyManager that returns the "actual targets" of
what this DependencyManager solves. This exposes an API to query not
just the targets, but the actual state that results from running the
transform. The actual targets are a superset of the targets that can
prevent unnecessary re-lowering.

Consider the situation of a dependency manager that has one transform
target, foo, and no current state. The flattened transform order that
results pulls in hundreds of other transforms that foo requires.
However, from the perspective of a user looking at the dependency
manager, it's "effect" is only running foo.

This matters when using a DependencyManager to solve a sub-problem. If
the sub-problem is too narrow, e.g., one transform foo, then the
hundreds of other transforms that may be run will be repeatedly run
over and over again because there's no API to query the effect of
solving a (targets, currentState) tuple.

This commit includes a small test that shows this issue.

This commit also removes a spurious addition of ResolveKinds before
ExpandWhensAndCheck.

For the provided test, the old order, without this PR is:

├── firrtlTests.options.ActualTargetsFixture$A
├── firrtlTests.options.ActualTargetsFixture$B
├── firrtlTests.options.ActualTargetsFixture$C
├── firrtl.options.PhaseManager
│   ├── firrtlTests.options.ActualTargetsFixture$A
│   └── firrtlTests.options.ActualTargetsFixture$B
├── firrtlTests.options.ActualTargetsFixture$D
├── firrtl.options.PhaseManager
│   └── firrtlTests.options.ActualTargetsFixture$A
└── firrtlTests.options.ActualTargetsFixture$E

The new order, with this PR is:

├── firrtlTests.options.ActualTargetsFixture$A
├── firrtlTests.options.ActualTargetsFixture$B
├── firrtlTests.options.ActualTargetsFixture$C
├── firrtl.options.PhaseManager
│   ├── firrtlTests.options.ActualTargetsFixture$A
│   └── firrtlTests.options.ActualTargetsFixture$B
├── firrtlTests.options.ActualTargetsFixture$D
└── firrtlTests.options.ActualTargetsFixture$E

The unnecessary re-lowering of A is removed.

This is related to the comment in #1784 (comment).

Note: doing this highlighted two lingering issues of using anonymous transforms in tests. Those are fixed in the first commit of this PR.

Contributor Checklist

• Did you add Scaladoc to every public function/method?
• [n/a] Did you update the FIRRTL spec to include every new feature/behavior?
• Did you add at least one test demonstrating the PR?
• Did you delete any extraneous printlns/debugging code?
• Did you specify the type of improvement?
• Did you state the API impact?
• Did you specify the code generation impact?
• Did you request a desired merge strategy?
• Did you add text to be included in the Release Notes for this change?

Type of Improvement

• performance improvement

API Impact

None.

Backend Code Generation Impact

None.

Desired Merge Strategy

• Squash: The PR will be squashed and merged (choose this if you have no preference.

Release Notes

• Improved transform scheduling

Reviewer Checklist (only modified by reviewer)

• Did you add the appropriate labels?
• Did you mark the proper milestone (1.2.x, 1.3.0, 1.4.0) ?
• Did you review?
• Did you check whether all relevant Contributor checkboxes have been checked?
• Did you mark as Please Merge?