Functionality to write ORSO dataset from combination of several runs

[jokasimr]

Currently the Orso provider looks for the NormalizedIofQ from the workflow and produces an ORSO dataset from that.
But often we want to put several curves in one ORSO file, either stitched curves or individual curves stacked in the ORSO file.

Some things are not clear and will have to be figured out:

• Does ORSO already have support for listing several reflectometry curves in the same file?
• How is the metadata from different runs combined?
• How is the Q-resolution from different runs combined?

It's possible that we will have to do a simple version for now and revisit this when we know more.

[paracini]

Let's start with putting multiple non-stitched angles measured on the same sample into one ORSO file
Some clarifications:

• ORSO support for multiple datasets in a single file: https://www.reflectometry.org/advanced_and_expert_level/file_format#multiple-data-sets-1
• MetadataThere are no strict indications on how to treat the metadata in the case of multiple datasets so there is flexibility. Let's start by having most information in the main header such as the identifier of each one of the multiple reduced reflectivity curves contained in the file and what they are (e.g. what sample rotation was used for each one in the case of measurements at different angles). This parameter can then be repeated as a single line header above the 4 columns of each dataset for readability
• Resolution resolution is reported for each individual curve. The main reason for keeping the datasets separate is the different q resolution of the datasets in the overlapping regions which should not be combined

[jl-wynen]

There is actually an explanation for how to encode metadata:

overwrite meta data

Below the separator line, metadata might be added. These overwrite the metadata supplied in > the initial main header (i.e. data set 2 does not know anything about the changes made for data set 1 but keeps any values from data set 0 (the header) which is not overwritten.

And note the example below this block that shows how multiple input files might be combined.

[paracini]

Thanks, I should read the links I post. In addition, there should be in the main header a short section that lists all the multiple measurements present in the dataset. Currently the main header has:
# data_files: raw data from sample
# - file: file name or identifier doi

Since these are then overwritten by the specific metadata above each dataset the main header should contain a list of identifiers of all the files present in the file.