support an option to run at the default max problemSpace

I plan on implementing this option (tonight) - it'd be great if you could provide quick feedback.

Given that each test has a very different meaning to the problem space (1d vs 2d, algorithm class) - it is difficult to scale something like the problem-space argument I provided in PR #18 across all tests.

What would the argument name be? --use-largest-default-problemspace ?

I need a capability such as this to reduce the amount of information coming from the tests to help give insights into each GPU or platforms strengths and weaknesses, that is all the smaller sizes would just lead to too many plots. The curent defaults inside were already well chosen sizes, so I think we should have a way to leverage them.

Review PR #18 and the existing handling of the problem-space argument and per-test defaults. Determine how the new option should select the largest existing default problem space, then verify that smaller test sizes are excluded and only the intended plots are produced.

Do mô hình lập chỉ mục viết ra từ nội dung của issue.