Changing default k-point density

[utf]

Did you see this paper: https://www.sciencedirect.com/science/article/abs/pii/S0927025619300813

Looking at figure 2, it seems like you should be able to define cut-off lengths that will give convergence for ~90 % of semiconductors and metals (using two separate cutoffs).

It might be nice for kgrid to default to these cut-offs, maybe with a --metal option to change to the default metal cutoff.

[ajjackson]

I'm not averse to playing with the defaults. I think the current value was essentially inherited from Graeme Watson.

So the proposal would be to

• change the default cutoff to whatever is recommended in that paper
• add a --metal option which sets a higher default, more suitable for metals
• do some argparse magic so --metal and --cutoff conflict