refactor: remove duplicated POD scientific modules

[VPRamon]

Closes #8

Summary

This refactor makes spod a service/application on top of the canonical Siderust 0.12 POD APIs rather than a second scientific POD toolkit.

Scientific consolidation

• Removed the duplicated local core, dynamics, estimation, observations, products, and qc implementations and their obsolete exports/tests.
• The service pipeline imports Siderust POD estimation, observation, force, product, QC, run-manifest, and propagation APIs directly.
• Replaced the deprecated finite-difference STM workaround with siderust::pod::propagation::VariationalPropagator; the initial state and identity STM are explicitly aligned with observation epoch zero.
• Removed unsupported synthetic carrier-ambiguity injection; the MVP estimates the six state components plus receiver clock.

Service boundaries and provenance

• CLI runs hash the actual YAML configuration with Siderust DatasetRef.
• REST jobs persist a deterministic serialized request.json in the run directory and hash that real file for manifest provenance.
• The current synthetic-only runner rejects every configured real-data input instead of hashing or recording data that the pipeline does not consume.
• Unsupported third_body: true configurations are rejected rather than silently running a TwoBody/J2-only model.
• Successful runs emit canonical Siderust RunManifest metadata with real SHA-256 configuration provenance, RFC3339 start/finish timestamps, deterministic references, and hashed output byte counts.
• spod::service no longer re-exports reusable scientific state types; internal callers import them directly from Siderust.

Cleanup

• Removed deprecated SIDERUST_POD_REST_BIND / SIDERUST_POD_REST_OUT compatibility.
• Removed legacy SPICE DAF/SPK parser re-export modules; low-level callers use siderust::formats::spice::{daf, spk} directly while spod::spice keeps only value-added kernel/provider functionality.
• Rewrote the active changelog, examples/doctest guidance, SPK documentation, and relevant ADRs for the post-consolidation architecture.
• Tightened the legacy namespace guard so obsolete project namespaces cannot return outside the explicitly historical workspace-reset ADR.
• Removed obsolete scientific dependencies in the preceding consolidation commit.

Tests and validation

• Added tests/service_synthetic.rs, covering RunConfig -> runner -> Siderust variational propagation/estimation -> products/QC -> RunManifest, including estimator output, required manifest fields, timestamps, configuration hash, output existence, and hash/byte-count consistency.
• The E2E test now asserts that the synthetic manifest contains exactly the configuration dataset as its input provenance.
• Added regression coverage that sp3, rinex_obs, rinex_nav, and antex are all rejected while real-data ingestion is unsupported.
• Added regression coverage for unsupported third-body configuration and for persisted REST request provenance (path, bytes, and SHA-256).
• CI validates formatting, checks, Clippy, feature combinations, docs/doctests, project guards, MSRV, coverage, Docker/REST/CLI smoke tests, and supply-chain policy.

The former standalone scientific API is intentionally not preserved through compatibility façades. Reusable POD science belongs upstream in Siderust; this PR keeps concrete service orchestration and external-format boundaries in spod.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 30f3fa2e59

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".