Fix selection bug in case of sorting / filtering

[jokasimr]

To select the correct rows from the data frames we must use Datagrid.get_visible_rows() when indexing to get the selected rows. The indices returned by Datgrid.selection refers to the indices in the visible part of the dataframe.

Depends on #148

[nvaytet]

Do you use the row_idx anywhere else in this function (and other functions below)?
If you don't, how about making a method named e.g. get_run_number() which would hide all the logic?
It appears that all methods are using the same code?

    def get_run_number(self, selections):
        row_idx = selections[0]['r1']
        return self.runs_table.get_visible_data().iloc[row_idx]['Run']

I don't know if you can even include the selections = self.runs_table.selections in there to reduce duplication further?

[jokasimr]

how about making a method named e.g. get_run_number() which would hide all the logic?

If you look at the other cases when we use selections (in the other classes in the module) you see that we don't always just get the first row from the selected region.

We could make a helper function, but I don't think it's that useful. The difference would be that instead of remembering to call grid.get_visible_data() we would have to remember to call helper_function_name(grid) when accessing data from a datagrid.

That said, we could do something like this:

def iloc_for_selected(grid):
    return grid.get_visible_data().iloc

and it would be used like

iloc_for_selected(self.runs_table)[row_idx]['Run']

[jokasimr]

Tried adding the helper function and I think that was a bit nicer.