feat: return null if no defaultValue is specified

[WikiRik]

Pull Request Checklist

Please make sure to review and check all of these items:

• Have you added new tests to prevent regressions?
• Does yarn test or yarn test-DIALECT pass with this change (including linting)?
• Is a documentation update included (if this change modifies existing APIs, or introduces new ones)?
• Did you update the typescript typings accordingly (if applicable)?
• Does the description below contain a link to an existing issue (Closes #[issue]) or a description of the issue you are solving?
• Did you follow the commit message conventions explained in CONTRIBUTING.md?

Description Of Change

This PR adds a defaultValue of null if there is no defaultValue specified when creating a model and the attribute allows null.
This fixes #14671.

@ephys I'm not so sure about this change. As you can see the unit tests fail because it also adds DEFAULT NULL to the SQL query. Is that wanted behaviour? This might even qualify as a breaking change

Todos

• See what to do about the failing tests

[ephys]

defaultValue could be 0, or '', a better check would be:

Suggested change

	if (attribute.allowNull !== false && !attribute.defaultValue) {
	if (attribute.allowNull !== false && attribute.defaultValue === undefined) {

It's also a good opportunity to normalize attribute.allowNull just before this so the test can be reduced to

Suggested change

	if (attribute.allowNull !== false && !attribute.defaultValue) {
	if (attribute.allowNull && attribute.defaultValue === undefined) {

[ephys]

As you can see the unit tests fail because it also adds DEFAULT NULL to the SQL query. Is that wanted behaviour? This might even qualify as a breaking change

The only element I was worried about was https://dev.mysql.com/doc/refman/8.0/en/timestamp-initialization.html, but after checking DEFAULT NULL is the default behavior for nullable columns there too

I think this is a safe change (but the tests need to be updated)

[WikiRik]

@ephys looking at the failed integration tests, I think it would be better to do this change after #14260 so that the impact of this on the tests is smaller. Do you agree?

[ephys]

Do we want to do #14260 in v7? I was thinking of postponing it to v8 to reduce how many impactful breaking changes we have, and maybe add the flag to change the default nullability in v7?

[ephys]

I think this PR can be undrafted once #14715 is complete, that other PR should implement the missing pieces

[WikiRik]

We still have an open issue connected to this, I'll close this for now and we can revisit with that issue later