Retire stale ROADMAP.md and move durable architecture to scoped docs

[daviburg]

Problem

ROADMAP.md is no longer a reliable planning source for the .NET SDK:

• v0.14.0-preview.1 ships 98 generated connector clients, while the roadmap still reports only two connectors with actions.
• It marks shipped clients such as Office 365 Users, Excel Online Business, Salesforce, DocuSign, and Event Hubs as not started or unevaluated.
• Its execution plan still includes completed work such as NuGet installation guidance and generated Microsoft connectors.
• It mixes .NET SDK work with Functions-extension, LSP UI, Python, Node.js, Java, connector-service, and conceptual-documentation ownership.
• The issue tracker and release changelog now provide more current, reviewable status.

Maintaining a second, manually synchronized status inventory is creating misleading guidance rather than planning value.

Proposed change

Retire the root roadmap instead of attempting another full status rewrite:

1. Move any durable trigger architecture or settled design decisions that exist only in ROADMAP.md into docs/triggers.md.
2. Move durable connection-setup architecture, if any is unique, into docs/connection-setup.md.
3. Delete connector rollout tables, dated execution checklists, success metrics, and cross-repository plans that are represented by shipped code, README inventory, or GitHub issues.
4. Delete ROADMAP.md and remove repository links to it.
5. Use GitHub issues, labels, and focused design documents as the planning source of truth.

Acceptance criteria

• No unique, still-valid architectural decision is lost.
• ROADMAP.md and all links to it are removed.
• README.md remains the shipped connector inventory.
• docs/triggers.md contains the current trigger status and architecture without dated implementation checklists.
• Open work is represented by scoped GitHub issues with owners or dependency labels.