docs: normalize remaining project naming in legacy docs

[ProfRandom92]

Summary:

• Normalize remaining human-facing project-name occurrences in docs/API_SURFACE.md and docs/research_positioning.md to CompText V7.
• Follow up on Gemini review feedback from PR docs: align legacy positioning with project source #158 after that PR was merged before the final naming cleanup landed.
• Keep the change documentation-only and limited to two already-discussed legacy docs.

Changed files:

• docs/API_SURFACE.md
• docs/research_positioning.md

Scope:

• Documentation-only.
• No runtime behavior changes.
• No validator changes.
• No fixture changes.
• No artifact changes.
• No evidence-index changes.
• No workflow changes.
• No package changes.
• No source-code changes.
• No README changes.

Non-goals confirmed:

• No agent framework behavior.
• No orchestration behavior.
• No LLM judge.
• No embeddings/vector search.
• No fuzzy semantic matching.
• No cloud dependency.
• No dashboard/SaaS behavior.
• No taxonomy changes.

Testing:

• Not run locally; docs-only follow-up.

[gemini-code-assist]

Code Review

This pull request standardizes the project name spelling to 'CompText V7' across the documentation files. A review comment suggests improving the readability of a comparison table in 'docs/research_positioning.md' by removing redundant project name repetitions in the boundary column, as the name is already defined in the header.

[gemini-code-assist]

The repetition of the project name in every cell of this column is redundant since it is already specified in the column header. Removing it would improve readability and align with the style used in other tables and lists in the documentation (e.g., the 'Not claimed' section on line 83 and the 'Stable API routes' table in docs/API_SURFACE.md).

Suggested change

	| RAG evaluation | Retrieval quality, answer grounding, citation coverage, or generated-answer quality. | CompText V7 does not retrieve documents or judge generated answers. It checks whether fixture-defined operational state survives compact/replay cycles. |
	| Vector memory | Embedding-based recall and similarity search over stored memories. | CompText V7 does not use embeddings or vector databases. It compares explicit fixture IDs, fields, attachments, and normalized values. |
	| KV-cache compression | Serving-layer efficiency for model attention/cache reuse. | CompText V7 does not optimize model internals or inference caches. It emits reviewable replay artifacts and field-survival metrics. |
	| Workflow orchestration | Durable execution, retries, scheduling, state machines, and tool execution. | CompText V7 does not run autonomous workflows. It validates whether replayed operational state still contains fixture-defined continuation requirements. |
	| Learned context compression | Model-learned summaries or compressed prompts optimized for downstream performance. | CompText V7 does not train or evaluate a learned compressor. It measures deterministic replay preservation under controlled fixtures. |
	| RAG evaluation | Retrieval quality, answer grounding, citation coverage, or generated-answer quality. | Does not retrieve documents or judge generated answers. It checks whether fixture-defined operational state survives compact/replay cycles. |
	| Vector memory | Embedding-based recall and similarity search over stored memories. | Does not use embeddings or vector databases. It compares explicit fixture IDs, fields, attachments, and normalized values. |
	| KV-cache compression | Serving-layer efficiency for model attention/cache reuse. | Does not optimize model internals or inference caches. It emits reviewable replay artifacts and field-survival metrics. |
	| Workflow orchestration | Durable execution, retries, scheduling, state machines, and tool execution. | Does not run autonomous workflows. It validates whether replayed operational state still contains fixture-defined continuation requirements. |
	| Learned context compression | Model-learned summaries or compressed prompts optimized for downstream performance. | Does not train or evaluate a learned compressor. It measures deterministic replay preservation under controlled fixtures. |