Add many-to-many relationship InvestigationSample to the schema

[EmilJunker]

This closes #231.

Here is a short summary of the changes:

• adds a new class InvestigationSample to the schema
  (representing a many-to-many relationship between Investigation and Sample)
• changes the unique constraint for the Sample table from investigation, name to just pid
  (adding a NOT NULL constraint for pid while also keeping the existing NOT NULL constraint for name)

Note that this is still very much a work in progress. A few points to note:

• The upgrade script initializes the value of the Sample.pid field with 'local:<id>' (reusing the id attribute) for all samples where the pid is currently NULL. However, in the current ICAT schema, the pid field has no unique constraint. This means that there could already be two samples with the same pid (at least theoretically). Or perhaps there could be several samples with the pid set to '' (an empty string). The upgrade script currently does not handle these cases (as they are very unlikely to occur in practice).

• The upgrade script also needs to move the existing one-to-many relationship between Investigation and Sample over to the new InvestigationSample table. Once this is done, the unique constraints on the Sample table are updated and the Sample.investigation_id column is dropped. I'm not sure if the current implementation of the upgrade script handles all of this correctly. In other words: thorough testing is needed.

[EmilJunker]

I'm closing this PR in favor of #294.