models: incorrect run number assignment after many restarts

[mdonadoni]

How to reproduce:

1. Execute one workflow (e.g. a demo example)
2. Restart the workflow ten times
3. The workflow runs will have the following run numbers:
   • workflow.1 (initial run)
   • workflow.1.1 (first restart)
   • workflow.1.2 (second restart)
   • ...
   • workflow.1.9 (ninth restart)
   • workflow.2 (tenth restart)

However, the tenth restart should be workflow.1.10.

At this point, it is also not possible to create a new workflow (reana-client create ...) with the same name. The workflow creation will fail given that workflow.2 is already being used:

(psycopg2.errors.UniqueViolation) duplicate key value violates unique constraint "_user_workflow_run_uc"
DETAIL:  Key (name, owner_id, run_number)=(workflow, 00000000-0000-0000-0000-000000000000, 2) already exists.

The cause of the issue is that run numbers are stored as floats and the new run numbers are generated in this way:

reana-db/reana_db/models.py

Line 566 in dd1b60e

return round(last_workflow.run_number + 0.1, 1)

One possible solution would be to store each run number as two integers and update the rest of the codebase accordingly.

[audrium]

Temporarily closed with #176. Reopening for the future to implement a different way how to store the run_number value in the DB

[audrium]

Created a new issue to solve the remaining limit of nine restarts #186