new hook: compare_identity_default

Question:

short of monkey-patching

# alembic.autogenerate.compare
def _compare_identity_default(...):

what is "the proper way" to override the logic of compare_identity_default?

https://github.com/sqlalchemy/alembic/blob/master/alembic/autogenerate/compare.py#L1022

The use case:

• database is multi-master postgres cluster
• ddl from alembic is replicated to all nodes via pglogical https://www.2ndquadrant.com/en/resources/pglogical/
• replication process on each cluster node intercepts ddl to inject its own per-node values for minvalue, maxvalue to ensure unique non-overlapping sequence range for identity for every node, like so:

-- this is substituted by ddl intercept on every node by values provided by postgres.conf file
... GENERATED ... AS IDENTITY ( ... MINVALUE %s  MAXVALUE %s  ... )

The requirement:

• have alembic ignore minvalue, maxvalue mismatch for identity columns
• during subsequent invocations of alembic revision --autogenerate
• depending on: database name, table name, column name

Start with alembic/autogenerate/compare.py around compare_identity_default and review how alembic revision --autogenerate invokes it. Determine an extension point that can use the database, table, and column names to ignore identity minvalue and maxvalue differences; done means the behavior is configurable without monkey-patching and works for subsequent autogeneration runs.