Resetting instrument geometry shouldn't delete the rest of the instrument

[Peter Peterson]

EditInstrumentGeometry should error out if not enough information is provided rather than quietly deleting the instrument.

[Nick Draper]

http://www.mantidproject.org/Release_Notes_2_2_Full_List

[Wenduo Zhou]

Make algorithm quit under incomplete information. Refs #5420.

Changeset: 81cf8eead16afc52a076a8e38b19efec4b6b4988

[Russell Taylor]

Input error checking looks stronger now.

Several comments:

• Unit tests for the failure cases would be a good idea.
• A negative PrimaryFlightPath looks to have a special meaning in the code, but this is not documented anywhere
• Most of the delete spec2indexmap statements (which are missing in several necessary places) could be avoided by localising the initial creation close to the one place where it's actually used and deleting right afterwards.
• vector::push_back (see lines 195-200) is inefficient and should be avoided if you know the size of the vector at the point of declaration (which you do).

[Wenduo Zhou]

Make algorithm quit under incomplete information. Refs #5420.

Changeset: 81cf8eead16afc52a076a8e38b19efec4b6b4988

[Stuart Campbell]

This ticket has been transferred to github issue ​6266