fix: add PolicyMfaFactors model to policy unions

[Meldiron]

What does this PR do?

Appwrite Cloud now returns an mfa-factors policy from GET /v1/project/policies (server model PolicyMFAFactors, added in appwrite/appwrite), but this SDK's unions predate it. As a result, every project.list_policies() call against Cloud currently crashes with:

AppwriteException: Unable to parse response into PolicyList: 23 validation errors for PolicyList
policies.10.PolicyPasswordDictionary.enabled
  Field required [type=missing, input_value={'$id': 'mfa-factors', ...}]
...

(caught by the Connect snippets QA in appwrite/vibes — all four python lanes fail; other SDKs parse leniently and are unaffected)

Changes, mirroring the generated style:

• New appwrite/models/policy_mfa_factors.py — PolicyMfaFactors with totp / email / phone / custom booleans, matching the server's PolicyMFAFactors response model.
• Added to the PolicyList.policies union and exported from appwrite.models.
• Added to the get_policy return union, docstring, and $id dispatch in services/project.py.
• Added MFA_FACTORS = "mfa-factors" to the ProjectPolicyId enum.

Not included: the update_mfa_factors_policy() service method (PATCH /v1/project/policies/mfa-factors) — the next spec regeneration will add it; this PR is the minimal fix for the parsing crash. If a regen is imminent, feel free to close this in favor of it — the vibes QA temporarily installs from this branch until a release containing the fix is on PyPI.

Test plan

PolicyList.model_validate({
    "total": 3,
    "policies": [
        {"$id": "password-strength", "min": 8, "uppercase": True, "lowercase": True, "number": True, "symbols": True},
        {"$id": "session-alert", "enabled": False},
        {"$id": "mfa-factors", "totp": True, "email": True, "phone": True, "custom": False},
    ],
})

parses successfully, with item 2 typed as PolicyMfaFactors; model_dump(by_alias=True) round-trips. Existing policy types still parse.

🤖 Generated with Claude Code

[greptile-apps]

Too many files changed for review (355 files, 200 file limit).

Bypass the limit by tagging @greptile-apps to review.