Allow sub-array ranges to end with 0 in order to include the last element

GREL provides an elegant way to extract a sub-array from an array using two indices, for example value[1,3], as shown in the documentation. Also very convenient is the ability to specify the second value as negative in order to count from the end, such as "internationalization"[1,-2] (also from the documentation). As it stands, however, this second style can't be used to extract a sub-array that includes the last element (i.e. one can't specify the last n elements) - for that you need to revert to using slice, or (in the case of a string) substring (or one could make a cumbersome use of the length function).

While these other functions will do the job well enough, I think it makes sense to simply allow 0 to be used as the second value in the range. For example "ABCD"[-2,0] to give "CD". I imagine a counter-argument might be that this would be the only case where that second value is not an actual array element (or rather, 0 could appear to refer to the first element) but since the element represented by that second value is always excluded from the subset (e.g. [1,3] does not include the [3] element), currently one can never sensibly use 0 there anyway - that is, there is no sub-array range, other than a trivial empty array, which can be described with [n,0]. To put it another way, I'm proposing that the 'count from end' logic, when used for the end of a range specification, should apply to values less than 1 rather than just negative values.

Although, as I mentioned, we're not lacking in alternatives at the moment, I keep finding myself having to turn to the documentation to remind myself of the right syntax for slice just to get around this. So I think the proposed feature would make this sub-array mechanism more elegant while still being logical.

Start by locating the GREL implementation for two-index range expressions and the tests covering positive and negative range endpoints. Verify the current behavior for examples such as "ABCD"[-2,0], then add coverage showing that an endpoint of 0 includes the final element while existing range behavior remains unchanged.