Add upper bound to pandas whenever it is a transitive dependency

[dbeatty10]

Proposal

To fully mitigate the blast radius reported in dbt-labs/dbt-core#9420, we should do something similar to #122 for all requirements.txt files that transitively include pandas.

More detail

pandas 2.2.0 has some breaking changes versus pandas 2.1.4.

These breaking changes can surface to users of dbt-databricks because pandas is a transitive dependency, and this was reported in dbt-labs/dbt-core#9420.

In order to mitigate the effects of dbt-labs/dbt-core#9420, we did #122 for dbt 1.5, 1.6, and 1.7.

We should do the same going back at least to dbt 1.3 9 (and maybe further).

The buggy behavior could theoretically go back at least to 1.3 since that is when dbt began enabling python models.

But there could be a transitive dependency on pandas via databricks-sql-connector that goes back further than that -- it just might not actually have an effect on dbt.