feat: fix multiple bugs related to sync() and indexes. Add Model.getIndexes()

[ephys]

Pull Request Checklist

• Have you added new tests to prevent regressions?
• Does yarn test or yarn test-DIALECT pass with this change (including linting)?
• Did you follow the commit message conventions explained in CONTRIBUTING.md?

Description Of Change

Closes #12889

This PR fixes the following bugs:

• DB2 forced columns that had a unique index to be non-null, this has been removed (you still need to make your column non-null, but we don't silently make that happen).
• When defining an index through the indexes option on a model, then calling sync(), the dialect would modify the model to mark all attributes that are part of the index as unique. (DB2 only)
• Calling sync({ alter: true }) with an attribute that has unique: true set would generate a query such as ALTER TABLE users ALTER COLUMN emails SET VARCHAR UNIQUE. DB2 and MSSQL do not allow UNIQUE to be specified there. This has been changed to use a separate CREATE INDEX query.
• SQLite supports specifying an index name during CREATE TABLE, but that name is ignored by sqlite. Causing subsequent sync() to try to recreate the index. The first sync does not try to create the index as part of CREATE TABLE anymore, but executes CREATE INDEX afterwards to create them (sqlite only, other dialects still do it during CREATE TABLE).
• This PR adds Model.getIndexes() which lists all indexes defined on the model (it regroups all indexes, the ones defined through ModelOptions.indexes, as well as the ones coming from ModelAttributeColumnOptions.unique)

This PR blocks #14572 (comment)

TODO

• Add tests for Duplicated unique constraints and indexes on sequelize.sync({ alter: true }) #12889
• Add sync(alter) tests for adding unique constraints after the first sync()

[ephys]

You're going to laugh. I've deadlocked my PRs.

One of the tests is failing because DB2 is recreating the table instead of altering it.
This is a DB2 bug I fixed in #14572
That PR is blocked because it relies on this one to fix DB2 bugs, and this one is blocked because it relies on that one.

This isn't a regression, so I'm requesting to force-merge this PR, and continue over at #14572

[WikiRik]

Can we temporarily disable the failing unit tests so only DB2 checks fail for this PR and re-enable them in the other PR? Not a fan of force merging

Will review tomorrow

[ephys]

Can we temporarily disable the failing unit tests

I fixed them, these weren't supposed to fail.
The only tests that were 'supposed' to fail were the DB2 integration ones, which funnily enough I cannot reproduce locally so there is more flakiness going on (still need to complete #14453)

Edit: looks like all remaining tests are flaky. I guess we're good to go

[sdepold]

I'm always surprised by how much you know about the library and in which corners you can stir :)

[ephys]

Let me handle the merging of this PR, I'll add a detailed message for the changelog

[github-actions]

🎉 This PR is included in version 7.0.0-alpha.14 🎉

The release is available on:

• npm package (@release-v7 dist-tag)
• GitHub release

Your semantic-release bot 📦🚀