Metric for remaining available database rows

[fabiante]

See #39 (comment)

Currently, the id columns of the purls table is an 32bit integer. A future version of persul will have to go for a 64bit integer. To avoid downtime due to reaching this limit, the system should implement a health check based on the remaining available row count in the purls table.

The idea is to check every X seconds how many remaining rows are available in the purls table. This is a metrics which should be observed using Prometheus or similar and alert operations once a threshold is reached.

This would then be the trigger for the application to upgrade the id column from a 32 bit to a 64 bit integer.