back to non-schematic 'sublocale' and 'interpretation' (despite df8fc0567a3d) for more potential parallelism;

help JVM to cope with large symbolic structures;

improved scheduling of forked proofs, based on elapsed time estimate (from last run via session log file);

read logs from failed sessions as well;
proper output base directory (which is two steps upwards);

recover timing information from old log files;
use session timing for queue ordering;
pass command timings to ML process (still unused);

suppress timing message in full PIDE protocol -- this is for batch build;

support for build passing timings from Scala to ML;

support for prescient timing information within command transactions;

emit command_timing properties into build log;
tuned;

generate Isar proof if Metis appears to be too slow