Graeme Winter

Run with ``` xia2 atom=Se read_all=0 trust_beam=1 \ image=/Volumes/Blue2/Data/i04-sethau-mad/SeThau_hrm_2_master.h5:1:900 \ image=/Volumes/Blue2/Data/i04-sethau-mad/SeThau_hrm_3_master.h5:1:900 \ image=/Volumes/Blue2/Data/i04-sethau-mad/SeThau_if_2_master.h5:1:900 \ image=/Volumes/Blue2/Data/i04-sethau-mad/SeThau_if_3_master.h5:1:900 \ image=/Volumes/Blue2/Data/i04-sethau-mad/SeThau_pk_2_master.h5:1:900 \ image=/Volumes/Blue2/Data/i04-sethau-mad/SeThau_pk_3_master.h5:1:900 ```

Suspect that this relates to the second splitting where we only get 3 sets of files not 6 - see #494 - which means that there are no data for...

OK, the `split_experiments` method of `DialsScalerHelper` knows nothing of the `by_wavelength` so expects (and encodes) 6 output files when only 3 appear here

Branch to `scale-multipass-mad` to investigate

To reproduce: ``` xia2 \ image=$(dials.data get -q fumarase)/12287_1_E1_001.img:1:30 \ image=$(dials.data get -q fumarase)/12287_1_E1_031.img:31:60 \ image=$(dials.data get -q fumarase)/12287_1_E2_001.img:1:30 \ image=$(dials.data get -q fumarase)/12287_1_E2_031.img:31:60 ```

Proposed test case: Take existing single wavelength xia2 test case, write new `xinfo` file which contains the data split into two logical wavelengths. Run xia2 on these. Verify that there...

Minimal test (to be improved) in `logical-wavelength` branch

Please could you show the content of the `H5-xia2.xinfo` file? I suspect you have found a corner case in the handling of resolution limits. When I ran `xia2 d_min=0.8 image=...`...

I have just tested with `d_min=` on the command line and xinfo input, without the `RESOLUTION` line, and it behaved as expected. I agree that `d_min=` on the command-line should...

FWIW I have verified that the `xinfo` `RESOLUTION` takes priority over the command-line `d_min=`