EPIC: issues with large number of small cuts

[abuts]

User was working with big number of small cuts and identified bunch of interconnected issues, which are the essence of this epic. They would probably separate into multiple smaller tickets

1. Incorrect experiment combine. Some parts will be sorted in bug(s?) in combine #1778 but some are related to 2. here. bug(s?) in combine #1778 assumes that files with the same run_id should contain same IX_experiment but in fact this should be checked and case of different IX_experiments dealt with. At least throw error for the beginning.

2. slow operations if loading/doing multiple small cuts and moving objects between containers. (hashes are recalculated at loagind and if there are a lot of small cuts loading). Claimed 7-fold deceleration. Solution -- Hashable class -- subclass of serializable. It would also allow fast comparison of hashable IX_experiments. Additional acceleration would be providing using mex-code for binning and transforming pixels into appropriate coordinate system.

• hashable class: hashable class  #1788
• prerequest for make instruments, samples and detectors hashable #1790 : proper comparison Serializable equality method (eq) should be replaced by overloaded equal_to_tol #1147
• make classes in containers hashable: make instruments, samples and detectors hashable #1790
• Enable full mexing options in cut #896 mex-ed binning. hashable (make instruments, samples and detectors hashable #1790) gives 3-fold acceleration, while remaining code in Horace-3 was mexed. Enable full mexing options in cut #896 written to enable this mexing in Horace-4

3. constant grows of unique_references container and its design. Design introduces additional implementation of singleton which is non-trivial to understand. Also adds all experiment's references to unique references container which for N small cuts containing M(n) references so N*M element grows for each operations with this bunch. Design and Grows -- separate but related issues.

• Prerequest for 3) -- standard form of singleton in unuque references containter Re Problems with unique references singleton implementation #1791

• bug in combine_experiment bug(s?) in combine #1778