docs: use SOLO_CACHE='default' and warn about LocMemCache (#157)

[AquaticAzelf]

Fixes #157

Problem: README.md:245 used SOLO_CACHE = 'local' which maps to LocMemCache. This is per-process (each worker has its own private cache) and is not memory-efficient — see Django docs: Local-memory caching. In multi-process production (gunicorn, uwsgi, etc.) this causes inconsistent singleton caching.

Solution: Change example to shared backend SOLO_CACHE = 'default' (Memcached) and add warning block:

Warning: LocMemCache is per-process (each process has its own private cache instance) and is not memory-efficient — see Django docs. Do not use SOLO_CACHE='local' in production with multiple processes/workers; use a shared backend (e.g. MemcachedCache/RedisCache) via SOLO_CACHE='default'.

Keep existing Caching will be disabled if set to None. sentence.

Verification (folder-local only, no global installs):

• python -m venv .venv inside clone C:\Users\Azelf\AppData\Local\Temp\opencode\django-solo-fix, pip install -e . (installed django-solo 2.5.1, Django 6.1)
• grep -n SOLO_CACHE README.md before: 245: SOLO_CACHE = 'local'; after: 245: SOLO_CACHE = 'default' plus warning line 248
• python manage.py test solo --settings=solo.tests.settings — 13 tests OK (exit 0)
• git diff shows 1 file changed, 3 insertions(+), 1 deletion(-)
• Repo stars verified ~966 (not zero), docs-only low risk.

Files: README.md only.