grid_max should return the last grid point rather than the upper bound of the interval for includeendpoint=false

[shinaoka]

If includeendpoint=false, grid_max should return the last point of the grid, rather than the supremum.

We could introduce grid_supremum returning b for $[a, b)$.

[shinaoka]

I got my head around this point. supremum is mathematically correct, but sounds unfamiliar. How about the following.

• Rename the attributes of DiscretizedGrid as grid_max -> upper_bound and grid_min to lower_bound.
• grid_max function returns the last grid point.
• We introduce upper_bound and lower_bound functions.

[shinaoka]

@rittermarc
I have a draft now. Does it make sense?
https://github.com/tensor4all/QuanticsGrids.jl/tree/7-invalid-grid_max-for-includeendpoint%3Dfalse