[#487] One SessionFactory per test class

[DavideD]

Fixes #487

Work on this with @blafond.

Now the test will create the factory once per test class. This saves some time with dbs like Db2 or CockroachDB.
The only issues I had is that some times tests get stuck but I couldn't figure out why. Refactoring some tests so that they now
use transaction solved the issues most of time.

If you do this, make sure that you recreate the factory after any test failure. Otherwise, the failure of one test method would have a cascading effect on all the other tests in that class (because its data won't be cleaned up).

@gavinking. do we really need to do this? The cleanDb method is called even when there is a failure. I suppose we could reset the factory if there is a problem deleting the entities but we don't have that many tests in a single class so I don't think it would be a problem. The reason I would prefer not to do it is because the only place I could do something like this is after the deleteEntities method has been called. I might thinkg of a better solution later.

For some reasons, this test doesn't work with Db2: https://github.com/hibernate/hibernate-reactive/compare/main...DavideD:487-sf-per-class?expand=1#diff-659762d6654e6fae801ebfcf7577499a86f6799bd9ce7948922726acc994fbab
Unless I recreate the factory each time.

I've also removed JUnit 5, in the end we don't need it and having both dependencies on the classpath might only causes issues.
I might add it back in the future if they add an extension that's now missing in vertx-junit-5

[gavinking]

do we really need to do this?

Well I think so, yes. I mean, I don't trust us to always write a cleanDb() method. Indeed, that is a total PITA to have to do.

[DavideD]

Well I think so, yes. I mean, I don't trust us to always write a cleanDb() method. Indeed, that is a total PITA to have to do.

@blafond is having a look to see if we can find the name of the entities and create a default deleteEntities that will work for most cases

[DavideD]

Something doesn't work as expected on the latest JDKs.
I've tried a quick experiment and noticed that the error message is different. Instead of containing the expected
Detected use of the reactive Session from a different Thread it says Session/EntityManager is closed (
https://github.com/hibernate/hibernate-reactive/pull/683/checks?check_run_id=2221537360#step:7:988)

Note that, because I wanted to see the error message, I've changed the assertion for that specific run. The assertion failing is actually:

hibernate-reactive/hibernate-reactive-core/src/test/java/org/hibernate/reactive/MultipleContextTest.java

Line 57 in 29733eb

testContext.assertTrue( e.getMessage().contains( ERROR_MESSAGE ) );

It would be interesting to know why there is a different error.
I will be on holiday until next week, if somebodys want to have a look at it is more than welcome, otherwise I will get back to it later.

[DavideD]

I think this is ready to be merged now. If anything because it fixes some existing tests

[blafond]

Looks good. Clean implementation and provides test classes option to manage DB entities life cycle.

[DavideD]

The build is still failing on CI sometimes. I'm having a look

[DavideD]

Merged #683

We can tweak it later when problems occur.