chore: Fix warnings for custom postgres_## cfg flags

[iamjpotts]

Resolves build warnings related to custom cfg conditions that are present in the code because they are set for various CI runs defined in a matrix in the GitHub workflow.

Example warning that disappears with this change:

warning: unexpected `cfg` condition name: `postgres_15`
   --> tests/postgres/types.rs:691:24
    |
691 | #[cfg(any(postgres_14, postgres_15))]
    |                        ^^^^^^^^^^^
    |
    = help: consider using a Cargo feature instead
    = help: or consider adding in `Cargo.toml` the `check-cfg` lint config for the lint:
             [lints.rust]
             unexpected_cfgs = { level = "warn", check-cfg = ['cfg(postgres_15)'] }
    = help: or consider adding `println!("cargo::rustc-check-cfg=cfg(postgres_15)");` to the top of the `build.rs`
    = note: see <https://doc.rust-lang.org/nightly/rustc/check-cfg/cargo-specifics.html> for more information about checking conditional configuration

This changes the postgres cfg conditions to be similar to how cfg conditions are setup for mariadb - version is a value of the postgres cfg parameter rather than having a separate parameter/condition for each postgres version.

Additional changes

Since Postgres 12 and prior are now EOL and CI was already adjusted In December to exclude Postgres 12 and earlier after Postgres 12 went EOL in November, some code comments and docstrings are adjusted to reflect that.

Docker compose file - version: 3 was removed since the version field has been deprecated for some time and generates a warning.

Github workflow:

• Job steps which had multiple docker commands are split into one step per docker command (easier to review logs)
• Steps which ran ssl auth tests are split into their own job - this prevents conflicts on host port 5432, which were happening even though the docker container was torn down (using docker compose rm -f did not help either)

#3879 will need modification after this is merged.

Does your PR solve an issue?

• reduces build warnings related to --cfg flags.
• eliminates warning related to deprecated docker compose version field

Is this a breaking change?

No - informs linter of an expected set of possible custom cfg parameter values, and adjusts test code, but not production code (other than comments)

[iamjpotts]

The CI failure is a transient database connection error:

error: error communicating with database: Connection reset by peer (os error 104)

I do not see a way to rerun the failed job.

[iamjpotts]

I do not see a way to rerun the failed job.

Resigned last commit and forced push to rerun jobs.

[iamjpotts]

Same transient failure as before, on same Postgres (13, async-std, native-tls) job:

Run cargo test --no-default-features --features any,postgres,macros,_unstable-all-types,runtime-async-std,tls-native-tls
warning: /home/runner/work/sqlx/sqlx/Cargo.toml: file `/home/runner/work/sqlx/sqlx/tests/sqlite/macros.rs` found to be present in multiple build targets:
  * `integration-test` target `sqlite-macros`
  * `integration-test` target `sqlite-unbundled-macros`
warning: /home/runner/work/sqlx/sqlx/sqlx-cli/Cargo.toml: `default-features` is ignored for sqlx, since `default-features` was not specified for `workspace.dependencies.sqlx`, this could become a hard error in the future
   Compiling sqlx v0.9.0-alpha.1 (/home/runner/work/sqlx/sqlx)
error: error communicating with database: Connection reset by peer (os error 104)
    --> tests/postgres/postgres.rs:1149:5
     |
1149 |     sqlx::query!("SELECT 1 AS one")
     |     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
     |
     = note: this error originates in the macro `$crate::sqlx_macros::expand_query` which comes from the expansion of the macro `sqlx::query` (in Nightly builds, run with -Z macro-backtrace for more info)

error: could not compile `sqlx` (test "postgres") due to 1 previous error
warning: build failed, waiting for other jobs to finish...
Error: Process completed with exit code 101.

[abonander]

It could be a race condition with Postgres starting up. I think we normally sleep a certain amount of time to make sure it's up first.

[iamjpotts]

Same failure, different postgres 13 job.

[iamjpotts]

I found two more places where the old cfg flags needed to get removed (docs).

[iamjpotts]

time="2025-08-16T01:17:41Z" level=warning msg="Found orphan containers ([postgres_13]) for this project. If you removed or renamed this service in your compose file, you can run this command with the --remove-orphans flag to clean it up."

This error from the step near the end that does docker compose -f after docker stop may be related.

[iamjpotts]

It could be a race condition with Postgres starting up. I think we normally sleep a certain amount of time to make sure it's up first.

@abonander the ultimate cause of the failure was a bind failure when docker was trying to use host port 5432 for the job step that stopped the non-ssl container and created the ssl container.

Job step with failure:

sqlx/.github/workflows/sqlx.yml

Lines 305 to 310 in 5f8fc6b

	# client SSL authentication
	- run: |
	docker stop postgres_${{ matrix.postgres }}
	docker compose -f tests/docker-compose.yml run -d -p 5432:5432 --name postgres_${{ matrix.postgres }}_client_ssl postgres_${{ matrix.postgres }}_client_ssl
	docker exec postgres_${{ matrix.postgres }}_client_ssl bash -c "until pg_isready; do sleep 1; done"

I've amended this pr to split the postgres job into two. That change could be a standalone pr if you wish.

[iamjpotts]

@abonander I'm done with my flurry of branch updates used to test CI changes.

I also cherry picked the last commit which is the fix into its own PR if you prefer that: #3977.