Handle units in return types of functions + lookups for parameters

[Nick Draper]

Add a result_unit parameter to the xml and if it is specified then we need to use it and the ConvertUnits code to convert the value through to the fitting unit.

we need to be able to handle these at least

• TOF and other units
• 1/TOF

Posibilities

1. possibly use muparser to evaulate the math string, but first work out the conversion factor between the base units. That way we should even be able to handle 1/(TOF)²
2. Handle the parsing yourself (not as good).

[Anders Markvardsen]

(In [4952]) Code for processing result-unit in LoadInstrument and added boost build flag to VS. Refs #1308

[Anders Markvardsen]

(In [4953]) Add processing for result-unit to FitParameter and add tests. Refs #1308

[Anders Markvardsen]

(In [5007]) Allow y-unit syntax for interpolation elements of IDF. Add this concept to Interpolation and testing this. Added additional user warnings in case users do not specify units recognisable by mantid Updated result-units in HRPD_Parameters.xml.

Refs #1308

[Russell Taylor]

(In [5009]) Revert qtiplot.pro changes, presumably checked-in in error. Re #1308.

[Anders Markvardsen]

(In [5721]) Simplification in preparation to handle parameter unit conversion. re #1308

[Anders Markvardsen]

(In [5789]) Implement rest for handling unit conversion for parameters. Added test and fitting BackToBackExponential to HRPD in different units gives the same result. Also updated wiki.

re #1308

[Anders Markvardsen]

(In [5790]) This might make it build on the build server (removing an API::). re #1308

[Anders Markvardsen]

(In [5793]) changed const double& to double (same or int) in argument list. maybe this will get the build to work?? re #1308.

[Roman Tolchenov]

(In [5795]) Linux build fix. re #1308

[Nick Draper]

fill in missing resolution

[Anders Markvardsen]

(In [5931]) Added unit to parameters for IkedaCarpenter function in GEM_Definition.xml.

Note that unit conversion is a bit more complicated for this function because we in that function calculate lambda values for each x-axis values. Hence in specifying the units for the IC parameters I have ignored units that refers to this lambda. This applies to two parameters. The true unit of Kappa is 1/[dSpacing^{2] however it is treated as dimensionless. The true unit of alpha1 is [TOF/dSpacing] however treated as it has unit TOF....}

Tested that I could fit IC function for a peak in workspace index 1 of Test/Nexus/focussedGEM38370_TOF.nxs and same starting width when fitting in both d-spacing and TOF.

Also changed warning "IkedaCarpenterPV function should ONLY be used when working with x-axis unit = TOF" to information: "IkedaCarpenterPV function is perhaps best used when working with x-axis unit = TOF"

(Further work on IC needed. Tickets will be generated.)

Refs #1308.

[Anders Markvardsen]

(In [5933]) Fix broken test. Refs #1308.

[Anders Markvardsen]

With reference to 3rd comment above and "(Further work on IC needed. Tickets will be generated.)". Tickets generated for this are #1546 and #1547.

[Martyn Gigg]

Verified in [5943]

Confirmed for GEM and IkedaCarpenter that fit works when in units other than TOF.

[Stuart Campbell]

This ticket has been transferred to github issue ​2155