Handle ±∞ in TimeSeries.add_timeseries()

[khaeru]

442 appears to be related to handling ±numpy.inf in input to TimeSeries.add_timeseries()—possibly on Oracle databases.

This PR adds test and fixes the issue by:

• Store the DB properties when the JDBCBackend is first created.
• Check the driver and for infinite values in input data.
• Raise an informative exception.

How to review

• View test_set_data_inf() in the diff.
• Try calling .add_timeseries() with infinite value when connected to a Oracle database via JDBCBackend gives the same kind of exception.

PR checklist

• [x] Continuous integration checks all ✅
• [x] Add or expand tests; coverage checks both ✅
• [x] Add, expand, or update documentation.
• [x] Update release notes.

[codecov[bot]]

Codecov Report

Merging #443 (2daa0cf) into main (0ac069e) will increase coverage by 0.0%. The diff coverage is 100.0%.

@@          Coverage Diff          @@
##            main    #443   +/-   ##
=====================================
  Coverage   98.5%   98.6%           
=====================================
  Files         42      42           
  Lines       4424    4441   +17     
=====================================
+ Hits        4362    4379   +17     
  Misses        62      62           

Impacted Files	Coverage Δ
ixmp/backend/jdbc.py	95.2% <100.0%> (+<0.1%)	:arrow_up:
ixmp/core/timeseries.py	98.7% <100.0%> (ø)
ixmp/tests/backend/test_jdbc.py	100.0% <100.0%> (ø)

[khaeru]

@Jihoon since you were able to confirm this when working against an Oracle database, I've asked for your review on the chance you can quickly re-do the steps you used to confirm, except with this branch.

If that's not possible LMK and I will re-request @iiasa/messageix-devs so it can be dispatched to someone else.

[khaeru]

If that's not possible LMK and I will re-request @iiasa/messageix-devs so it can be dispatched to someone else.

Since Jihoon is currently having lots of fun with NGFS, I changed the request.