Peripheral tables cause issues: deletes, field name uniqueness, etc.

[CBroz1]

By peripheral tables, I refer to tables like IntervalList and AnalysisNwbfile. These tables ...

• Have many foreign key references throughout various pipelines
• Centralize a data type (e.g., interval, file path)

Pros:

1. Centralize processes that manage their type (e.g., cleanup, interval list operations, fetch_nwb)
2. Facilitate operations that compare instances, though I'm not aware of current cases.

Cons:

1. Not captured by cascading deletes, requiring the use of operations like cleanup (see #948)
2. Result in redundancy (see #778) requiring workarounds like cautious_insert
3. Result in keys managing multiple values for these data types (see #961), requiring renaming
4. Cause issues tracking an entry through a pipeline, providing false parent paths for RestrGraph

An alternative design would replace foreign key references with fields and centralize enforcement of the data type via a utility.

Converting existing pipelines is impractical.