Add a draft THREAT_MODEL.md + SECURITY.md and link it from AGENTS.md

[potiuk]

This is a v1 draft threat model for the OpenDAL PMC to review — please
correct, reject, or discuss as needed. The maintainer is the decision-maker.

Lands a draft THREAT_MODEL.md for apache/opendal-oli and wires the
AGENTS.md → SECURITY.md → THREAT_MODEL.md discoverability chain.

oli is a CLI veneer over the OpenDAL core, so the model defers the
data-access trust boundaries (credentials-in-transit, per-backend auth,
hostile-backend handling, backend-side paths) to the core model in
apache/opendal#7641, and covers the CLI-specific surface: the
config/profile + credential source, argument/URI handling, local-FS
interaction for cp, and not echoing secrets in output.

• THREAT_MODEL.md (new) — the CLI-specific draft (defers to core).
• SECURITY.md (new) — routes reports to the ASF security process; points at the model.
• AGENTS.md (new) — Security section linking SECURITY.md.

Produced from the repo's public artifacts following the rubric at
https://gist.github.com/scovetta/2dc9a0695c7cbcc32e23799e00d2ced3. Claims are
tagged (documented) / (inferred); the inferred ones are collected as §14
open questions — a one-line confirm/correct each is enough.

Context: the ASF Security team is preparing OpenDAL for an automated agentic
security scan being piloted by the team; a discoverable model is the one hard
pre-flight gate. One of the 4 in-scope OpenDAL repos (core #7641,
reqsign #761, opendal-go-services, and this). Questions / pushback welcome.