feat: heap size control for recon-util

[c-dilks]

@gavalian says it's too small; this is for testing a high-stats run in clas12-validation with larger max heap size.

[c-dilks]

testing with JeffersonLab/clas12-validation#177

we know 1500m fails because JeffersonLab/clas12-validation#172

[c-dilks]

allocating 3GB speeds clas12-validation reconstruction up quite a bit: JeffersonLab/clas12-validation#177

So.... let's make heap allocation control a feature, and leave the default the same.

[baltzell]

Let's pursue a more generic approach that supports passing any arbitrary JVM options, for all these frontend scripts. Could push it to OptionParser, or use the old-school -- separator for CLI arguments in the shell wrapers, or ....

See #604

[c-dilks]

Let's pursue a more generic approach that supports passing any arbitrary JVM options, for all these frontend scripts. Could push it to OptionParser, or use the old-school -- separator for CLI arguments in the shell wrapers, or ....

See #604

Fully agree, as discussed. Comment linked to #604 description.