LonelyCat124

This is dependent on #1551 and will appear as all the code from that PR is here for now.

Ok, looking back at this again today, and found a big oversight in my `handle_binary_op` strategy, which I'm unclear on the way to fix. My assumption had been that if...

I fixed the previous issue, and have run into another oversight I need to fix. A normal use case might be something like: ``` do i = 1, 320 do...

I'm probably going to put design notes for the new ideas I have in this PR now. Operation order: 1. Find all parent loops inside the same parallel region. Its...

`handle_binary_op` at the moment does the following things, and is called only when a binary operation is found inside an array index: 1. If the binary operation is not an...

`handle_read_only_ref` at the moment does the following things: 1. For `(ArrayReference, ArrayOfStructuresReference)`: At the moment, the initial implementation checks whether this ArrayReference was private in the parent parallel. This can...

There are some uncovered loc in this PR right now, but they are implemented in #1551 so will be fixed when that is merged and this is rebased to master.

Ok - I tested this with NemoLite2D and found a variety of issues/bugs to fix. 1. StructureReference need to only add the outermost member to a clause, as it should...

1 should be handled for most structure references, however its notably not handled correctly when those structures are inside the declared loop's start/stop/step conditions, so I will fix that first.

Ok, a lot of puzzling today and I made a bit of progress and hit some potential hurdles. One thing I'm realising quickly is that since kernels use calls to...