Matrix 0-indexed

[jacobpake]

This PR makes matrices 0-indexed (for consistency with lists), updating the internal and library functions.

[| (i, j) for (i, j) in (n, m) |]

The above produces a n x m matrix as before, but i and j here will be 0..n-1 and 0..m-1 rather than 1..n and 1..m (is this what we want?).

dims will return the size as before.

Internals diff
Library diff

Most tests were straightforward to update but there are two notable cases that could use another set of eyes:

array.fld - the values in the matrix are initialised from the indices, so the matrix (and test output) now contains different values (is this right?)

compute-dtw.fld - output is different, either similar to above or I may have made a mistake in test/fluid/lib/dtw.fld (this was a bit delicate)

Fixes #790

[rolyp]

Thanks @jacobpake -- array.fld looks as expected. compute-dtw.fld I'm less sure about -- I'll take a look this morning.

[jacobpake]

compute-dtw.fld I'm less sure about -- I'll take a look this morning.

Thanks - I struggled to understand the calculateDTW functions - perhaps it could be clearer when / why we offset indices and add other documentation.

[jacobpake]

@rolyp I made the changes to range we discussed here.

I change ListEnum to use a new range_inclusive. Let me know if it should keep using range (and have the + 1 in exprFwd/exprBwd) instead or if these changes should happen in a separate PR.

[rolyp]

Looks great, just some minor suggestions re. the new exprBwd case for [e1 .. e2].