compiler: Make par-tile unboundedness customisable

[FabioLuporini]

This is related to the recent "x loop running on the host" issue that u @mloubout patched across OSS and PRO

I need a different fix (in PRO) because currently otherwise these two par-tiles for any 2-kernel Operator would generate different code!

• par-tile = (32, 4)
• par-tile = ((32, 4), (32, 4))

The latter would create an inner x loop of block size 4; the former a full loop spanning the whole dimension

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 84.13%. Comparing base (1069c63) to head (d85c9dd).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3040      +/-   ##
==========================================
+ Coverage   83.99%   84.13%   +0.14%     
==========================================
  Files         258      258              
  Lines       56422    56450      +28     
  Branches     4815     4817       +2     
==========================================
+ Hits        47389    47492     +103     
+ Misses       8203     8135      -68     
+ Partials      830      823       -7     

Flag	Coverage Δ
pytest-gpu-aomp-amdgpuX	68.66% <100.00%> (-0.01%)	⬇️
pytest-gpu-gcc-	78.82% <33.33%> (+0.01%)	⬆️
pytest-gpu-icx-	78.75% <33.33%> (+0.01%)	⬆️
pytest-gpu-nvc-nvidiaX	69.19% <100.00%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.