Write scaffolded code for solana bitget solver

[AryanGodara]

Description

Wire in the config, the CLI subcommand and the Dex variant for the new Bitget Solana solver. The swap adapter comes in the next PR, so up to this PR bitget answers every order with no solution.

Changes

• dex::Error: Jupiter's error enum moves up to the dex module, so the solve loop and Quote no longer name Jupiter's type.
• Make the config generic over the engine's [dex] table, loaded per subcommand.
• Add the Bitget config with endpoint, api-key, api-secret and slippage-bps. The struct has no Debug, so the secret can't reach the logs.
• Dex::Bitget with a stub swap that returns NotFound, and a bitget subcommand.
• Add an example config for the solver.

How to test

• CI (this PR is a no-op at runtime).

Closes BE-252

[linear-code]

BE-252

[AryanGodara]

this is done in next PR

[claude]

Claude finished @AryanGodara's task in 1m 49s —— View job

---

Review: Bitget solver scaffold

Reviewed the full diff against origin/main. This is a clean, well-structured scaffold — config made generic over the [dex] table, Error hoisted from jupiter to the dex module, Dex::Bitget stub, CLI subcommand, and run.rs de-duplicated so the serving path is shared. Test coverage for the new config (parse + required-credentials) is a nice touch, and omitting Debug on BitgetConfig to keep the secret out of logs is the right call.

No blocking issues. One non-blocking note posted inline:

• config.rs:94 — the endpoint trailing-slash requirement is enforced only by a comment; a missing slash will silently resolve to the wrong URL once Url::join lands. Worth normalizing/validating in the constructor in the follow-up (BE-253).

Everything else (the dex.rs:11 TODO) is already acknowledged as next-PR work. LGTM. ✅
· aryan/be-252-bitget-scaffold

[claude]

Non-blocking / for the next PR: correctness here relies entirely on a prose comment ("needs the trailing slash"). Because Url::join("swap") replaces the last path segment when the base lacks a trailing slash, a deploy config of .../swapx/pro (no slash) would silently resolve to .../swapx/swap rather than .../swapx/pro/swap — a wrong endpoint with no error. Worth validating/normalizing the trailing slash in the Bitget constructor when the swap adapter lands (BE-253), rather than trusting operators to read the comment.