Batch sampling utility

[kjappelbaum]

Expected Behavior

In dispersant_screener we had the batch sampling argument in the pal function. There should be an easy way to do the same here.

Actual Behavior

Here, I didn't implement it as it simply means that the sample function needs to be run n times.

Options

• refactor the sampling function to also take the number of samples
• add a utility/wrapper function

[kjappelbaum]

There's actually a design issue that I forgot:

• we decided to update the mask array with the sampled points only in the update_train_set() function, which makes sense to me as before that the sampling didn't really happen. This is, a sample without subsequent measurement does not really help us and we do not know if only some or all objectives have been measured
• for this reason, I suggest we implement exclude_idx in the _get_max_wt function such that we can exclude some indices in the case of batch sampling