drainer: reduce memory usage when restart

[you06]

What problem does this PR solve?

close #978

When drainer restart, it'll load all DDL history from TiDB cluster, this will cause a large memory usage.

What is changed and how it works?

This PR change the DDL history load strategy, load only recently DDLs into drainer, for the stale DDLs, load from a snapshot schema.

Check List

Tests

• Manual test (add detailed scripts or steps below)

Release note

• Reduce memory usage when drainer restart

[sre-bot]

No release note, Please follow https://github.com/pingcap/community/blob/master/contributors/release-note-checker.md

[sre-bot]

No release note, Please follow https://github.com/pingcap/community/blob/master/contributors/release-note-checker.md

[you06]

This is a draft implementation.

Because the relationship between job.BinlogInfo.SchemaVersion and job.ID is unknowable, drainer still to load all DDL history to check whether it's schema version is newer than current binlog, the most work we do makes non sense, and it'll cost a long time loading history.

[you06]

/run-all-tests

[you06]

/run-integration-test

[you06]

/run-all-tests

[ti-chi-bot]

@you06: PR needs rebase.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}