(cherry picked from commit 20eef6dc4a4867b80bd5999f679d78a1561ac896)

(cherry picked from commit f673b84acbcf524704636766320eab8ef4ac7049)

(cherry picked from commit 821a0490122a5f1369ecaadfad3d4376ef92c29e)

the difference of using or not KERNELs in histogram
accumulation

I remembered why I was using find rather than ==.  It should now work
for the problematic cases though

(cherry picked from commit e5679dad)

added the options to set spacing instead of nbin and
the option to translate the feses with the minimum set
to zero.

This also fix #72.

More work is needed to make FuncSumHills fully compatible
with COLVAR files (i.e. adding the possibility of doing
histogram only on selected collective variables/fields instead
of using all the fields).