[initialisation step] Use of a temporary database during initialisation step

[servoz]

Following numerous discussions, the last of which related to the use of the Input_filter brick in all cases #243, it seems necessary to change the database indexing at the initialisation stage.

• We no longer want to index the database for i/o at initialisation time. Instead, the database should be copied at the beginning of the initialisation step, for example by copying the COLLECTION_CURRENT collection to COLLECTION_TEMP. This would solve the problem of initialising several different pipelines in several tabs, which can potentially lead to inconsistencies #150 (COLLECTION_TEMP would be overwritten every time the initialisation step starts).

• The indexing of the i/o would be done in this temporary database to allow the user to see the result of the initialisation.

• To do this at the end of the initialisation there would be an automatic opening of a window allowing the visualisation of this temporary database (something close to what we can observe in the data browser, but maybe a little bit simpler!).

• Currently we have interesting tags in the COLLECTION_BRICK collection like Init or Init Time. How can we not lose them if we overwrite the temporary database at each initialisation? We could imagine to create a temporary database for each tab and to overwrite the corresponding database each time we initialise in a tab. So at run time we could retrieve the temporary database (all or only the data we are interested in) to the tab where we run the pipeline.

*For more details, please see #243