ENH: `eval` should log computed metrics to screen

[NickleDave]

To get an initial sense of performance before loading .csv.

Currently the logger just reports that eval metrics were saved, as shown below

$ vak eval data/configs/gy6or6_eval.toml 
Logging results to results/gy6or6_eval
loading spect scaler from path: results/gy6or6_train/results_220315_122127/StandardizeSpect
loading labelmap from path: results/gy6or6_train/results_220315_122127/labelmap.json
creating dataset for evaluation from: data/prep/has_notmat_prep_220315_123036.csv
running evaluation for model: TweetyNet
Loading checkpoint from:
results/gy6or6_train/results_220315_122127/TweetyNet/checkpoints/max-val-acc-checkpoint.pt 
batch 206 / 207: 100%|███████████████████████████████████████████| 207/207 [00:06<00:00, 33.76it/s]
saving csv with evaluation metrics at: results/gy6or6_eval/eval_TweetyNet_220315_130531.csv