Remove upper-binding for SQLAlchemy

[potiuk]

There was a problem with custom classes for SQLAlchemy that
prevented it to work on MySQL. This PR removes the SQLAlchemy
upper binding.

---

^ Add meaningful description above

Read the Pull Request Guidelines for more information.
In case of fundamental code changes, an Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in a newsfragment file, named {pr_number}.significant.rst, in newsfragments.

[potiuk]

Let's see if the problem has been fixed in the meantime

[potiuk]

Hey @ephraimbuddy @dstandish - I got the tests for MSSQL workiig for latest SQLalachemy but I am not sure if the solution to the failing tetst were good.

There were a number tests added recently to those db migration scripts (good!) and apparently when I migrated to latest sqlalchemy the previous problem mentioned in #21272 is solved but this one tests comparing the schema with the ORM information failled. I am not sure if I fixed it correctly, the root cause of this is likely this one:

https://github.com/apache/airflow/blob/main/airflow/migrations/versions/0093_2_2_0_taskinstance_keyed_to_dagrun.py#L278

I believe the CASCADE modification was the problem. Both "diffs" that I ignored had "CASCADE" ondelete, where we are definitely setting NO_ACTION for MsSQL due to the https://docs.microsoft.com/en-us/sql/relational-databases/errors-events/mssqlserver-1785-database-engine-error?view=sql-server-ver15

So I guess for some reason our "diff" script expects CASCADE but gets NO_ACTION,

I just ignored the offending differences - similarly to others tests, but maybe this is an indication of some other problem we have (or maybe our diff is not robust enough to account for differences between the dialects?)

Any insight and comments (and maybe approval if I did it correctly) is appreciated.

[potiuk]

All Green.