docs: RFC for recurring scheduled payments via delegated authorization

[Davidemulo]

Summary

Issue #347 is an expert-difficulty, multi-week epic whose own acceptance criteria requires "a maintainer-approved RFC with an explicit threat model before implementation" as the first gate. This PR delivers that RFC only — no contract code, keeper service, or SDK changes — per the issue's own stated process. Implementation should follow as a separate, later PR once this design is reviewed.

docs/rfc-recurring-payments.md proposes:

• A Mandate lifecycle (create_mandate/execute_mandate/revoke_mandate/get_mandate) grounded in the existing SplitterContract's actual pay()/require_auth() conventions and id-counter pattern (read from the real contract source, not assumed).
• Reusing the token's existing SEP-41 approve/transfer_from allowance mechanism for the "may not over-pull" guarantee, rather than inventing custom signature-delegation primitives — the allowance bounds the lifetime total, the mandate's own on-chain interval/runs-completed counters bound the rate. execute_mandate takes no caller-supplied timing/amount data, so a malicious or buggy keeper cannot spoof "it's due."
• A threat-model table addressing early/over-execution, revocation finality, expiry, race conditions between concurrent keepers, and a griefing analysis for both withheld and malicious-but-harmless permissionless execution.
• An explicit "skip, not catch-up" recommendation for missed intervals, with reasoning.
• Open questions flagged for maintainer decision rather than silently assumed: allowance revocation UX after a mandate is revoked, controller/delegation scope, keeper fee responsibility, whether a due-list view function is worth adding, and the unbounded-mandate allowance ceiling.

Closes #347 (the RFC deliverable; full implementation is a follow-up gated on this review per the issue's own criteria)

Test plan

• Maintainer review of the proposed design and threat model
• Explicit decisions on the open questions in §5 before implementation begins