Add a "stale workflow" scanner/fixer impl, to remove data beyond reasonable retention

[Groxx]

For a variety of reasons, we have workflows sitting in some of our databases far beyond any reasonable time range.
E.g. a workflow completed months ago in a domain with 7 day retention -> unfortunately still retained.

We've been gradually finding and fixing problems that cause this, but the old data needs to be cleaned up
at some point. They can sometimes cause problems, particularly if they appear to be "running".

So this is a conservative scanner + fixer for workflows that meet these criteria.
It does not handle all states, nor does it support domains configured with >200 day retention (hardcoded but
trivially changeable, as a sanity check), but it's catching a large amount internally and seems safe to share
and run in more locations.

---

Since building and verifying this required figuring out the scanner/fixer code in detail,
and that proved to be a significant effort, this also includes three major additions on top
of the stale scanner:

• there have been a variety of renames and public -> private changes to make naming consistent
  • this helped discover a significant issue with the current execution scanner, so I'm keeping it
• concrete execution fixer invariants can now be individually enabled and disabled via dynamic config
• a new readme.md file that attempts to summarize the learnings, and how to config and verify scanner/fixer.

[coveralls]

Pull Request Test Coverage Report for Build 01899039-7b38-4151-afff-08a098661ffd

• 273 of 495 (55.15%) changed or added relevant lines in 8 files are covered.
• 45 unchanged lines in 13 files lost coverage.
• Overall coverage increased (+0.07%) to 57.115%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
common/reconciliation/invariant/collection_enumer.go	0	1	0.0%
service/worker/scanner/executions/current_execution.go	0	1	0.0%
service/worker/scanner/executions/types.go	0	5	0.0%
service/worker/scanner/executions/concrete_execution.go	0	7	0.0%
tools/cli/adminDBCleanCommand.go	0	8	0.0%
tools/cli/adminDBScanCommand.go	0	8	0.0%
common/reconciliation/invariant/staleWorkflow.go	266	458	58.08%

Files with Coverage Reduction	New Missed Lines	%
common/reconciliation/invariant/collection_enumer.go	1	0%
service/worker/scanner/executions/concrete_execution.go	1	36.36%
common/persistence/sql/sqlplugin/postgres/db.go	2	85.0%
common/task/fifoTaskScheduler.go	2	84.54%
common/task/weightedRoundRobinTaskScheduler.go	2	89.12%
common/util.go	2	53.44%
service/history/handler.go	2	47.28%
service/matching/taskListManager.go	3	80.89%
common/types/history.go	4	45.35%
service/history/historyEngine.go	4	68.79%

Totals
Change from base Build 01898906-4fcf-4fd0-9203-586f0c269abc:	0.07%
Covered Lines:	87519
Relevant Lines:	153234

---

💛 - Coveralls

[davidporter-id-au]

LGTM.

Overall the testing you did certainly makes me a lot more confident in the overall approach, though looked extremely annoying.

If you'd done the work already I'd suggest not doing the backwards-compat, but no strong feelings that it's there, though I would be awary introducing further complexity or working on the assumption or just set to terminate-if-running