1125 nexus writer tests
Closes issue #1125 (closed)
Fix issue 1149 (manual closing after testing at ID19)
- add writer tests
- variable-length scans and virtual datasets: resize before dataset creation + dummy data
- lima: fix issue with dataset uri (was pointing to the NXentry instead of the dataset)
- add code profiling (time and memory) to the writer
This adds significant time to the CI pipeline. We may need to reduce the number of tests. Currently I test each scan with
- configurable writer, with data policy, default options
- configurable writer, with data policy, alternative options
- configurable writer, without data policy, default options
- base writer, with data policy, default options
- base writer, with data policy, alternative options
- base writer, without data policy, default options
The "alternative options" refer to non-default writer options like allowing external EDF's etc. Then I test following scans:
- ct: 0D (time)
- timescan: 1D variable length (time)
- loopscan: 1D fixed length (time)
- ascan: 1D fixed length (motors)
- amesh: nD fixed length (motors)
All these combinations cause different writing behaviour so they are all relevant. But if we need to reduce tests I can test ct with all options and the others with the default option only (configurable writer, with data policy, default options).
Edited by Wout De Nolf