Implement more granular exception handling in update entity definitions processing

Problem/Motivation

The automatic entity schema update system is quite powerful. Simple changes to schemas of entities and fields are handled automatically when database updates are run. This allows new indexes to be added, or old ones to be changed or deleted, etc, all without custom update hooks for each change.

The problem however is that as soon as a single update cannot be processed (any detected schema change to storage for a field with data, for instance), no further automatic updates are processed.

This is problematic in that all of the awesomeness of the system is lost as soon as one change throws an exception.

Proposed resolution

Iterate through each change and catch exceptions, then throw the exception at the end. This will allow non-problematic updates to run, and isolate the failing ones.

Remaining tasks

User interface changes

API changes

Data model changes