Replace hard-wired inclusions in meta-panel.controller

[brianritchie1312]

meta-panel.controller uses a hard-wired list of inclusions for each entityType. This means that the meta-panel query always includes the same entities, not all of which may be required; it also means that the set of inclusions must be extended if new entity references are required (as happened recently with Study from Investigation).

The code extracts variable names from the meta-tab fields already; so it should be possible to replace the hard-wired .include()s by including the discovered variable names.

I am slightly hesitant because there may be other reasons why particular inclusions have been hard-wired; but this may be worth considering.

[brianritchie1312]

I have tried replacing the hard-wired includes by including the extracted variable names; but it does not work. As a simple counter-example: for LILS investigations, the only extracted variable name is 'investigationUserPivot'. The hard-wired list does not include this, but does include 'user'. Replacing the hard-wired list with only 'investigationUserPivot' leads to an AngularJS "digest iterations" runaway. I am not sure why this happens, but it discourages me from trying further at the moment.

[brianritchie1312]

Fixes #217 (which I've just rediscovered!)

[brianritchie1312]

I have found a counterexample which does not work with this change: a field expression such as:

datafileParameter[entity.type.valueType=='STRING'].stringValue

results in a red alert TypeError: entity.type is undefined. This is because though the new code detects the reference to datafileParameter and generates an inclusion for it, it does not detect the qualified expression, and cannot know that it should include datafileParameterType instead. The original code has the latter inclusion hard-wired (and it works).

Determining the correct set of inclusions from the configured metatab fields appears to be more difficult than anticipated. It is probably necessary to reuse or adapt the rather involved code used in the row calculations. I am not certain this is worthwhile.

Given that this counterexample is drawn from a production configuration (it is used by ISIS to display a metatab that shows all of a Datafile's parameters) I think that the branch that claims to fix this issue is not ready to be pulled into the master - and may never be! I am marking this issue as wontfix.