[script.*] Implement F1-score a priori where possible; Improve plotting of precision-recall curves with better legends, more consistence; Remove redundant plotting code; Update tests

[checkpointer] Make it flexible to directory renames; Fix saving of trainlog in case we are interrupted

[docs,scripts] Separate evaluation script; Improve trainer engine code; Reset network model names to match resource names

[script/binseg] In evalpred, allow user to specify the prediction map extension, but defaults to .png