Improve own test coverage and remove brittle expected_mutants_for_own_source_tree

[sourcefrog]

This brittle test was hiding some mutants that were not otherwise well tested: coverage is better with the additions in this PR.

Also, this test needed to be updated on many commits, which was tedious and in hindsight a bad smell.

Fixes #295

[sourcefrog]

There are now some timeouts in 93a850f:

src/outcome.rs:224:9: replace ScenarioOutcome::check_or_build_failed -> bool with false
src/outcome.rs:231:9: replace ScenarioOutcome::mutant_caught -> bool with false
src/outcome.rs:232:34: replace == with != in ScenarioOutcome::mutant_caught
src/process.rs:50:9: replace Process::run -> Result<ProcessStatus> with Err(::anyhow::anyhow!("mutated!"))
src/process.rs:129:9: replace Process::terminate -> Result<()> with Err(::anyhow::anyhow!("mutated!"))
src/process.rs:216:9: replace ProcessStatus::is_failure -> bool with false
src/scenario.rs:28:9: replace Scenario::is_mutant -> bool with false

(possibly more)

They seem to all originate in tree_with_symlink in build_dir.rs...

Perhaps this is because they're all running in the source tree, and having trouble locking files there?

[sourcefrog]

Trying to bisect for the change that started this:

git bisect start
# bad: [ace8b702e6e899df3218e8cf878604fe340d2153] Unviable mutation only needs a check build
git bisect bad ace8b702e6e899df3218e8cf878604fe340d2153
# good: [75a504cbf41017a8f22a9ec7d654428be4c2b610] Don't fail fast on mutants matrix (#297)
git bisect good 75a504cbf41017a8f22a9ec7d654428be4c2b610
# good: [b75f14491ec90a270521825e338008f760d1f367] Split out and improve tests for --check
git bisect good b75f14491ec90a270521825e338008f760d1f367

[sourcefrog]

Apparently one cause of the hangs was sourcefrog/nutmeg#16, a cute deadlock in Nutmeg, avoided for now by not sending trace messages while rendering progress.