Add test APIs to wait for a document to be saved in the background

Functional tests that check something saved in the background (e.g. by an editor's auto-save) have to wait for it, and today they do so with getDriver() calls in the test class (raising the driver timeout, polling with waitUntilCondition), which breaks the page-object boundary: the test shouldn't need to know how long to wait or how to poll.

Add TestUtils.RestTestUtils#waitForVersionChange(reference, fromVersion, timeout): wait for the version of a document to differ from a given one, whatever changes it (a save, a rollback, a version deletion), polling it over REST (so it works whichever browser tab, if any, changes it), and return the new version. It throws a TimeoutException otherwise, which also lets a test check that no new version is created.

Example: the BlockNote auto-save tests (AutoSaveIT). They use this API, plus page-object methods that wait for the auto-save with timeouts derived from the configured interval (BlockNoteEditor#waitForAutoSave() and BlockNoteEditor#isAutoSavedWithinNextInterval()), and no longer call getDriver().