feat(solana-solvers): report uniform clearing prices in engine solutions

[squadgazzz]

Description

The engine wire reports one executed amount per trade and nothing else, so the driver cannot know the trade's other leg and currently answers the autopilot with a zero placeholder that the ranking math rejects (discussed on #4759). The EVM engine wire solves this with uniform clearing prices per solution, and the Jupiter engine already holds both numbers: the quote's in and out amounts.

This adds the EVM-shaped prices map to the engine's solution DTO, oriented like the EVM solvers build it: the sell mint maps to the amount bought, the buy mint to the amount sold. The driver ignores unknown fields today, so this lands independently, and the driver-side consumption (deriving the counterpart leg and dropping the zero placeholder) follows in the driver's own PR.

Changes

• Engine solutions carry prices (mint to amount, decimal strings on the wire), populated from the executed swap's two amounts

How to test

Updated unit tests.

[claude]

Claude finished @squadgazzz's task in 1m 44s —— View job

---

PR Review

• Read changed file & PR diff
• Analyze correctness of prices mapping and wire format
• Check test coverage
• Post findings

The prices mapping itself looks correct and matches the EVM orientation described (sell mint → amount bought, buy mint → amount sold), and the SameMint/ZeroAmount guards ensure two distinct, non-zero entries.

One issue found — the wire_format_is_stable test will now fail. prices is serialized unconditionally (no skip_serializing_if), but that test's exact assert_eq! on the full JSON wasn't updated to include the new key. Details and a suggested fix are in the inline comment.

I couldn't run cargo nextest/cargo check here (sandbox blocks compilation), so this is from static analysis — please confirm the test result locally.