fix(config): use named constants for fallbacks, remove dead code, fix stale doc ref (#673, #674, #675, #678)

[trakshan-mishra]

Description

Bundles four config-module cleanups, all good-first-issues in oracle/src/config.rs and oracle/src/network_config.rs. Each is small, independent, and addresses the exact acceptance criteria in its issue.

#675 — bind_addr fallback duplicates the literal instead of reusing DEFAULT_BIND_ADDR

collect_or_default! for bind_addr used "0.0.0.0:8080".parse().unwrap() as its fallback, while every other tunable in the file reuses its DEFAULT_* constant. A future edit to DEFAULT_BIND_ADDR would silently desync from the fallback with no test catching it.

Fix: replaced the literal with DEFAULT_BIND_ADDR.parse().unwrap() and added a unit test (config_from_lookup_bind_addr_defaults_to_constant) asserting Config::from_lookup resolves to DEFAULT_BIND_ADDR when BIND_ADDR is unset.

#674 — ENV_KEY constant is unused; the literal "PRICE_FEED_CONFIG" is duplicated at the lookup

ENV_KEY was declared as the single source of truth for the env-var name, but the actual lookup at line 204 hardcoded the literal string. If the var name were ever renamed, ENV_KEY would silently go stale while the real lookups kept working off the duplicated literal.

Fix: replaced the literal at the lookup site with ENV_KEY. (The remaining PRICE_FEED_CONFIG references in shared/config/src/lib.rs are user-facing error messages and doc comments in a separate crate — not lookups — so they stay as literal strings describing the var name to users.)

#673 — MAINNET_RPC_URL is defined but never referenced

MAINNET_RPC_URL had no call site anywhere in the workspace. On Network::Mainnet, config.rs calls required(&mut lookup, "STELLAR_RPC_URL") unconditionally — there is no fallback path. This is intentional: the test config_from_lookup_requires_explicit_mainnet_rpc asserts that mainnet forces explicit RPC configuration (a safety measure). Wiring the constant as a fallback would break that test and the intentional design.

Fix: removed the dead constant rather than wiring it in, so the code stops implying a code path that doesn't exist. MAINNET_PASSPHRASE (which is used) is retained.

#678 — validate_strkey() doc comment cites an unrelated issue #3

The doc comment claimed CRC16/keypair-derivation validation is "wired with the keeper in #3", but #3 is an unrelated closed data-store rent estimation feature. No tracking issue exists for the CRC16/derivation gap.

Fix: reworded the doc comment to accurately describe the limitation without the stale issue pointer: full keypair verification happens in the keeper runtime, not at config load.

Type of change

• 🐛 Bug fix (config correctness + dead code + stale docs)

Verification

cargo fmt --all                          # clean
cargo clippy --all-targets -- -D warnings # 0 warnings
cargo test --workspace                   # 296+ tests, 0 failures

The new test config_from_lookup_bind_addr_defaults_to_constant passes, and the existing config_from_lookup_requires_explicit_mainnet_rpc test still passes (confirming the MAINNET_RPC_URL removal doesn't affect the intentional mainnet-requires-explicit-RPC behavior).

Related issues

Closes #673, #674, #675, #678