(At least 5 more places that weren't deleting the returned map.)

The BinaryOperationTable was not deleted at the end of the operation
causing a memory leak. The shared_ptr will take care of this properly.
Refs #7994

(At least 5 more places that weren't deleting the returned map.)

Added a pair of braces near BinaryOperation.cpp line 376 to stop an if being ambiguous

Refs #4489

BinaryOperation::checkCompatibility no longer uses a stringstream and just throws the message reported by BinaryOperation::checkSizeCompatibility
Some error messages were missing some punctuation.
One error message was re-worded.

Refs #4489

BinaryOperation::checkSizeCompatiblity and overridden versions in child
classes have been changed to return std::string rather than bool so that a
useful error message can be passed back down the call stack before finally
being used in throwing std::invalid_argument in BinaryOperation::checkCompatibility.
A return of true is now a return of "" (empty string). A return of false is
now a string of length 1 or more containing an error about why the
algorithm couldn't complete. Functions that used the bool return before
have been changed to use checkSizeCompatiblity(lhs,rhs).empty as that is
the equivelent of the old boolean return.
Removed a load of commented-out code from Multiply.cpp.
Indented WeightedMean.cpp and WeightedMean.h as per the guidelines.

Refs #4489

It is redundant alongside the ISpectrum objects and was only being used
in tests and in one place in the ISIS SANS GUI (for which I've put in what
I believe to be equivalent code, though there's no test to verify it (too
much logic in the GUI layer).
This also resolves part of #6198.

only MDBoxFlatTree may still need to be modified for this ticket.

The MVS2012 crashes on building BinaryOperation.cpp in release with debug information  mode. To get access to this mode I have to refactor binary operation code

This reverts commit 649d9f5c.

...as those on the Mac (see commit 6b4fcc91).

The problem looked to be due to the fact that the Intel compiler
evaluates function arguments L->R (out other compilers do it R->L). If
the LHS & output workspaces were the same one, the references obtained
by the readY/E calls could be invalidated by other threads, whereas
with R->L evaluation the dataY/E calls would have ensured the readY/E
gave the right one.
In fact, the tests that showed this up are rather artificial as it's
extremely unlikely that data will be shared between Y & E vectors
(unlike X vectors), but it's better to make sure it's correct.

Back to previous (before 6bde1e7a of Oct 2010) and, I believe, more
correct way of deciding whether we need to create a new output workspace.

Touches nearly everything unfortunately but is necessary.

It can cause problems with workspace groups.

The HFIR system test randomly failed on RHEL last night. Helgrind
flagged up a problem with the ParameterMap access again. This will probably
affect the performance but correctness is of higher importance.

Add a copy constructor to Instrument. Oh, and go on a const-correctness hunt that led this commit to touch slightly more than just the one class. Re #1944.

When masked _bins_ are propagated from an input to an output workspace the y-values shouldn't be modified re #3382

- Refs #3203: Implemented ISpectrum. It is the base class of EventList and Histogram and contains the spectrum number and list of detector IDs for a given spectrum.
- ISpectrum return X/Y/E vectors. EventList, e.g. calculates the histogram on the fly.
- Moved the EventWorkspace MRU list of histograms to be contained within the EventList; better code separation. Made a EventWorkspaceMRU class to encapsulate this more easily.
- Workspaces must implement getSpectrum (const/not const) to return a ISpectrum object. dataX(i), etc. methods on MatrixWorkspaces no longer need to be overridden since they call getSpectrum()
- Got rid of EventWorkspace::getEventListatPixelID() and doneLoadingData()
- replaceSpectraMap() methods offers backwards-compatibility with SpectraDetectorMaps by setting the specNo and detector list when called. Must be called AFTER creating all the spectra in a workspace.
- Refs #3318: maskWorkspaceIndex() implemented as part of ISpectrum.

Replace uses of the SpectraDetectorMap with the new interface. Note that mutableSpectraMap has now gone in favour of using MatrixWorkspace::replaceSpectraMap with a brand new implentation. Re #3067

Should fix 32 bit windows, probably at the cost of a warning but I'm not on the right platform to take a chance. Re #3008.