Introduce step cache to reduce load on deduplicate

[benjie]

Description

Deduplicate is expensive; this new step cache enables the duplicate steps to not be created in the first place, reducing the load on deduplicate. By carefully controlling the way in which steps can be cached (and when these caches need to be cleared) we can enable this functionality safely without risking the complexities that would occur with more naive approaches.

Performance impact

Should reduce planning time for most queries, however the act of writing to the cache does incur some performance penalty so it's certainly possible that some queries will be slower than before. I expect them to be faster in the vast majority of cases.

Security impact

Reduced memory usage (and time usage) during planning should reduce security threats slightly.

Checklist

• My code matches the project's code style and yarn lint:fix passes.
• I've added tests for the new feature, and yarn test passes.
• I have detailed the new feature in the relevant documentation.
• I have added this feature to 'Pending' in the RELEASE_NOTES.md file (if one exists).
• If this is a breaking change I've explained why.

[changeset-bot]

🦋 Changeset detected

Latest commit: ee30542

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 15 packages

Name	Type
postgraphile	Patch
@dataplan/pg	Patch
grafast	Patch
graphile-utils	Patch
pgl	Patch
graphile	Patch
graphile-build-pg	Patch
@localrepo/grafast-bench	Patch
@dataplan/json	Patch
@grafserv/persisted	Patch
grafserv	Patch
ruru-components	Patch
@localrepo/grafast-website	Patch
graphile-build	Patch
graphile-export	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[benjie]

This reduces the planning time of @slaskis' reproduction from ~6.25s to ~1.26s - i.e. it's now 5x as fast.

[benjie]

(The actual solution to @slaskis' issue is #2060; but this fix is welcome nonetheless.)