Meta: Apex as a Celestia-backed Developer Storage Platform

## Summary

Tracking issue for exposing Celestia blob storage through familiar developer interfaces.

The core idea: Apex already indexes and caches Celestia blobs in SQLite. By adding a write path and a generic blob abstraction, we can layer multiple storage APIs on top.

## Prerequisites

- [ ] #44 — Blob Submission Pipeline (write path — keyring, nonce, tx lifecycle)

## Storage APIs

- [ ] #40 — S3-Compatible Object Storage API
- [ ] #41 — Git Remote Storage via Apex
- [ ] #42 — Celestia-backed NPM Registry

## Architecture

```
Protocol-specific API (S3 / Git / npm)
↓
Object ↔ Blob chunking + metadata
↓
SQLite index/cache
↓
Blob Submission Pipeline (#44)
↓
Celestia DA
```

## Dependency graph

```
#43 (this — storage platform)
├── #44 (blob submission pipeline)
│ ├── #4 (tx client decision)
│ ├── #9 (keyring/signing)
│ ├── #5 (multi-account)
│ ├── #8 (nonce management)
│ ├── #18 (rate limiting)
│ └── #19 (tx confirmation)
├── #40 (S3 API)
├── #41 (Git remote)
└── #42 (NPM registry)
```

The key enabler is a common content-addressed blob storage abstraction that handles chunking, reassembly, and metadata tracking. Once that and the submission pipeline exist, each protocol adapter becomes a relatively thin translation layer.