chore(templates): move the API starters to PostgreSQL 18

[Bccorb]

Closes #41.

seamless-cli moved to PostgreSQL 18 (fells-code/seamless-cli#164), so a project from seamless init was naming two majors in the same directory: postgres:18 in the CLI-generated compose file and postgres:16-alpine in the API starter copied in beside it.

The tag was not the whole change

Bumping just the image breaks the starter. PostgreSQL 18+ images store data in a major-versioned subdirectory (/var/lib/postgresql/18/docker), and the mount has to move up a level to match. With the old /var/lib/postgresql/data mount the container treats the volume as unused and refuses to start, restart-looping on:

Error: in 18+, these Docker images are configured to store database data in a
       format which is compatible with "pg_ctlcluster" ...
       Counter to that, there appears to be PostgreSQL data in:
         /var/lib/postgresql/data (unused mount/volume)

See docker-library/postgres#1259. So this PR is two changes, not one:

	Before	After
image	postgres:16-alpine	postgres:18-alpine
mount	pgdata:/var/lib/postgresql/data	pgdata:/var/lib/postgresql

A comment on the mount records why, since the path now looks wrong to anyone who knows the old convention.

Verified, not assumed

I stood the express starter's db service up on a fresh volume both ways:

• image bumped, mount left alone → restarting, the error above in a loop.
• image bumped, mount moved → running (healthy), PostgreSQL 18.4 on aarch64-unknown-linux-musl, and PGDATA=/var/lib/postgresql/18/docker resolving inside the mounted volume, so data actually persists.

npm run validate passes.

Blast radius

None for existing projects. Starter files are copied in at scaffold time, so a project keeps the compose file it was created with. Only newly scaffolded projects get 18, on a fresh volume.

Follow-ups this uncovered

The mount-path change applies to every other postgres pin in the org, and none of the issues I filed mentioned it, because I did not know until I tested this one. Notably seamless-cli's own generated compose has the same defect and is already merged (fells-code/seamless-cli#164 pins postgres:18 with a /var/lib/postgresql/data mount). Fixing that is next; its CI missed it because the verify harness's postgres has no volume mount at all.

After this merges: tag a templates release, then bump SEAMLESS_TEMPLATES_REF in seamless-cli/src/core/images.ts.