Reporting a few bugs and relevant fixes

[brandonjwk]

When using eclasses, inference engine chooses wrong CPDs to add to joint potential
BNT\inference\static@jtree_inf_engine\enter_evidence.m
ln 53: e = bnet.rep_of_eclass(bnet.equiv_class(n));
BNT\general\compute_joint_pot.m
ln 12: e = bnet.rep_of_eclass(bnet.equiv_class(i));
BNT\general\convert_dbn_CPDs_to_pots.m
ln 17: e = bnet.rep_of_eclass(bnet.equiv_class(n, 1));
ln 24: e = bnet.rep_of_eclass(bnet.equiv_class(n, 2));

Using global joint inference engine 'exclude' parameter creates jpot with too large a domain
BNT\inference\static@global_joint_inf_engine\enter_evidence.m
ln 35: [engine.jpot] = compute_joint_pot(bnet, mysetdiff(1:N, exclude), evidence);

Rename strsplit to avoid conflicts with MATLAB built-in strsplit
KPMtools\strsplit.m >> strsplitKPM.m

[TigerZhao007]

Rename strsplit to avoid conflicts with MATLAB built-in strsplit
KPMtools\strsplit.m >> strsplitKPM.m

Thanks man, help me a lot.