Draft: rename modules for clearer naming (interfaces_* split, _index family, import/ingest)

Context

Follow-up to #112 and the discussion started there: shreejaykurhade picked up "[P3, medium] Change inconsistent module names" and audited import-style inconsistencies (which became #298 / #299-#306). @gvanrossum's request in that thread was specific:

I think we must make a plan for what renames we want, rather than letting Claude pick them... @shreejaykurhade Would you like to start with a draft list of modules to rename and propose new names? I can then review the list and we can quickly come to an agreement that way.

That draft never landed. This issue is a first attempt at it, scoped to two concrete, defensible cases rather than an exhaustive rename of everything — explicitly a draft for discussion, not a decision. Per gvanrossum's stated preference, the actual choice of names should be made by the team, not by an LLM; this is meant to give that discussion a concrete starting point, backed by an actual read of each module's contents.

1. The interfaces_*.py split: names don't match contents

knowpro/interfaces.py re-exports five submodules. I read what's actually defined in each. Two of them contain zero Protocol/interface definitions — they're pure data-type modules mislabeled as "interfaces":

interfaces_search.py and interfaces_serialization.py are pure data/schema modules with no interfaces in them at all — the "interfaces_" prefix actively misleads about their content. interfaces_serialization.py is especially confusing sitting next to serialization.py (the actual serialize/deserialize logic module) — same near-identical name, very different purpose (one is wire-format TypedDict shapes, the other is the code that uses them).

Draft proposal:

2. Module naming consistency: the _index family

The original P3 TODO ("Change inconsistent module names") is most visible here. messageindex.py and timestampindex.py spell their subject out in full; five siblings abbreviate it, each differently:

Draft proposal (matching each module's own class name, in the same fused-no-underscore style as the two reference modules):

• secindex.py → secondaryindex.py
• propindex.py → propertyindex.py
• reltermsindex.py → relatedtermsindex.py
• semrefindex.py → semanticrefindex.py
• textlocindex.py → textlocationindex.py

Open question for the team: fused (relatedtermsindex.py, matching existing messageindex.py/timestampindex.py) vs. underscored (related_terms_index.py, matching most other multi-word modules like email_message.py, universal_message.py). Both conventions currently coexist in the codebase; this rename is a chance to pick one for the _index family specifically.

3. Smaller, easy case: import vs. ingest

Three modules do the same job (parse a source format, produce ConversationMessages) but disagree on the verb:

• emails/email_import.py — function import_emails_from_dir
• podcasts/podcast_ingest.py — function ingest_podcast
• transcripts/transcript_ingest.py — function ingest_vtt_transcript

Draft proposal: standardize on ingest (2-of-3 majority) — rename email_import.py → email_ingest.py, import_emails_from_dir → ingest_emails_from_dir.

Non-goals

This is a starting draft, not a complete audit of every module name in the tree — it covers the two clearest, most defensible cases. No renames have been implemented; this issue exists purely to give the team something concrete to react to, per gvanrossum's request in #112.