feat: Databricks grants

[eakmanrq]

Unique nuances of Databricks for grants:

• When doing Unity Catalog (UC) granting for groups, which the engine adapter assumes UC, the groups need to be defined at the account level. When creating groups with SQL they are done at the workspace level which I think means at the hive catalog level. It would require making API calls to create roles at the account level. As a result I used pre-created roles at the account level to simplify things but if this becomes a problem we could update to do the API calls.
• Databricks doesn't support the standard UPDATE/INSERT permissions so I had to parameterize these to allow the dialects change what is used
• When querying the information schema Databricks would return inherited privileges which couldn't even be controlled through explicit grants. For example on the test account we have ALL_PRIVILEGES given by default and this would be returned and then "revoked" but the revoke would be a no-op since it wasn't explicitly granted in the first place. Therefore we just ignore inherited grants.

Confirmed CI tests for Snow/Databricks pass.

[newtonapple]

I assume the reason we lower case everything is because we are assuming Databricks users will be using Unity catalog? Is that always the case? I know some deployments still use hive metastore. If I remember correctly, it table names etc. can be case sensitive there.

[eakmanrq]

The engine adapter itself assumes Unity Catalog. So we don't need to worry about hive metastore compatibility.