Refactor PoldiFitPeaks2D to use EstimatePeakErrors

I tested the changes running this slightly modified usage example, which puts each of the workspaces into a WorkspaceGroup first. Without these changes the two resulting WorkspaceGroups contain the same workspace (the first one is overwritten), after the changes this does not happen, as described in the ticket.

```
trans1 = Load('INTER00013463')
trans2 = Load('INTER00013464')

trans1_wav = CreateTransmissionWorkspaceAuto(trans1)
trans2_wav = CreateTransmissionWorkspaceAuto(trans2)

wsGroup1 = GroupWorkspaces("trans1_wav")
wsGroup2 = GroupWorkspaces("trans2_wav")

stitched_wav1, y = Stitch1DMany("wsGroup1,wsGroup2", params='1, 0.02, 17', UseManualScaleFactor=True, ManualScaleFactor=0.85)
stitched_wav2, y = Stitch1DMany("wsGroup1,wsGroup2", params='1, 0.05, 17', UseManualScaleFactor=True, ManualScaleFactor=3.85)
```

FABADA increment

Port CutMD to C++

Test adaptive radius integration

IPython > 2.1 also works

Memory leak in LogParserTest

Horace style for cutmd

The actual order of the columns doesn't matter, so don't require an
arbitrary order.

Good starting values for time limits

Add class RemoteJobManagerFactory (+ ComputeResourceInfo)

bug in load isis nexus monitors

The unit and doc tests for this algorithm pass, both on the build servers as well as locally. The documentation reads well, all the links are working. I did not know FITS at all, now it's clear what kind of data are loaded by the algorithm and what the output workspaces contain.

Log plot precision.