WandB: Add metric logging support on eval end and epoch end

[arnavgarg1]

Our current callback implementation only logs the dataset, but not any metrics.

Confirmed that this branch works if you also downgrade wandb to 0.12.10. We should spend time fixing our integration/making it compatible with the latest version of wandb, but for now, this fix should suffice.

FYI, it's not clear what the issue is with the latest version (no code errors, just doesn't push metrics up to wandb). Here's the slack thread for context: https://ludwig-ai.slack.com/archives/C01PN6M2RSM/p1693941105546379

[github-actions]

Unit Test Results

       6 files  ±       0         6 suites  ±0   1h 54m 16s ⏱️ + 25m 10s
2 822 tests  -        4  2 785 ✔️  -        5  12 💤 ±  0  25 ❌ +1 
8 431 runs  +5 562  8 374 ✔️ +5 550  31 💤 +10  26 ❌ +2 

For more details on these failures, see this check.

Results for commit cc7929e. ± Comparison against base commit e1d023e.