Batch edit improvements

[sharadsw]

From #5417 (review)

1. Currently, there is no way to select specific rows and batch edit just them. This might be a requirement, maybe not. There is a work around to do so. The current implementation supports batch edit out of a record set. So, the user can:

• Make a query.
• Select the records, make record set out of it.
• Query out of the record set, and batch edit on those results.

2. Data sorting. Currently, if you have one table (so no relationships), sorting should always (of multiple columns) be respected (UX guys: test this). If you have relationships, and you also sort on some of the relationship's columns, it's essentially arbitrary (no guarantees regarding what user sees). From the code perspective, it is not arbitrary (see if you can figure those places out). I know that we don't support relationships, but this point actually applies regardless, since the same procedure is applied, just that we ignore relationships in upload plan. This might UX issue, a way to work around will be to sort in the workbench instead (in the spreadsheet).

3. Potential performance issues, as mentioned in a below comment in stored_queries/views.py. IIRC, I once tried batch editing andy's entire voucher collection (45k CO records with determination and cataloger) and it took 1.2 minutes to construct the dataset (I think that's too slow - I deferred profiling it for later, maybe need to that).

[sharadsw]

Improvements related to code:

• Check if predicates.py can be simplified (Batch Edit: Support for editing basic fields #5417 (comment))
• Use yield_per to fetch rows (Batch Edit: Support for editing basic fields #5417 (comment))