Abins: trouble restarting with a lower number of quantum events

[ajjackson]

Describe the bug The Abins algorithm caches some of its results and will attempt to re-use them if a similar calculation is run. An intended use-case of this is to calculate with a low quantum-order, and then re-run with a higher quantum order. The existing data should not be recalculated.

A bug is encountered when doing the reverse while autoconvolution is enabled. (Autoconvolution adds orders 2-10 if the main calculation is at order 1, and adds orders 3-10 if the main calculation is up to order 2.)

To Reproduce Run the Abins algorithm with any phonon data (VibrationalOrPhononFile) and corresponding value of AbInitioProgram. Set QuantumOrderEventsNumber to 2 and check Autoconvolution. Then run Abins again, changing the output workspace name and setting QuantumOrderEventsNumber to 1.

This raises an error message, showing the dictionary key 'order_2'.

Expected behavior Abins should read the existing order-1 data, recalculate the convolution orders (starting at order 2) and terminate normally, creating new workspaces.

Platform/Version (please complete the following information): I've mostly experienced this on Linux development builds but it has been there for a while and is assumed to be cross-platform.