Reconciliation improvements needed

[ehuelsmann]

After getting back up to speed and having studied both #6122 and #6183, I think #6183 can be addressed after we finish this one (which is spawned off #6122). What seems to be missing in this one is:

• [x] A final decision on what it means for the reconciliation if a single payment has multiple sources in its journal lines
  Solution: Add a field listing the payment id associated with the reconciliation row (the addition of this field is a 1.10 enhancement).
• [ ] A test case which tests reconciliation of payments and GL lines, all without a Source indicator (expected test result: separate groups between payments and GL; each payment its own group; each GL journal line its own group)
• [ ] A test case which tests reconciliation of GL lines and payments, with different source indicators between GL and payments (and between payments) -- with a single source per payment
• [ ] A test case which tests reconcilation of GL lines and payments where some GL lines have the same source as payments
  • [x] where the GL lines are added in the same run as the payments (expected outcome: ...)
  • [x] where the gl lines are added in a later run than the payments (expected outcome: ...)
  • [ ] where the payments are added in a later run than the GL lines (expected outcome: ...)
• [x] A test case which tests reconciliation of payments with multiple sources per payment

All of the above tested per payment/GL date, meaning that we need to ensure that sources of different dates don't mix into a single reconciliation line.

Are there any further test cases to be considered?

Originally posted by @ehuelsmann in https://github.com/ledgersmb/LedgerSMB/pull/6122#issuecomment-970703840