Create a CorrectByTransmission algorithm

[Nick Draper]

#Missing CorrectByTransmission algorithm to do the following

#Load transmission data
#Load direct data
#handle conversion of both to instrument transmission mode
#convert to wavelength and rebin
# transM3/transM2 = trans
# directM3/directM2 = direct
# correction = trans/direct
#rebin trans to data
# data/trans

[Russell Taylor]

(In [1864]) Adding an algorithm for linear least squares fitting. Uses GSL. Re #375.

[Russell Taylor]

(In [1866]) Windows doesn't seem to like using the WorkspaceFactory in the test. Re #375.

[Russell Taylor]

(In [1869]) Need to account possible zero errors. Re #375.

[Russell Taylor]

(In [1878]) First version of transmission correction for LOQ/SANS2D. Basically works, but still has some bells and whistles to be added. Re #375.

[Russell Taylor]

(In [1886]) Some corrections. Re #375.

[Russell Taylor]

(In [1888]) Update script. Re #375.

[Russell Taylor]

Need to take account of the errors in the fit.

[Russell Taylor]

(In [1952]) Take account of errors in fit to transmission curve. Re #375.

[Russell Taylor]

(In [1954]) Avoid negative errors. Re #375.

[Russell Taylor]

(In [2024]) Add test for CalculateTransmission. Also replace some dodgy hard-coding in the algorithm itself. Re #375.

[Russell Taylor]

(In [2025]) Re #375.

[Russell Taylor]

(In [2027]) Try the usual Windows test failure suspects. Re #375.

[Stuart Campbell]

This ticket has been transferred to github issue ​1223