feat(github): per-repo commit enumeration for full history backfill

Adds a new github-repo EventSource that enumerates all repos via
/user/repos and walks each repo's /commits?author= endpoint, which
has no 1000-result cap unlike the Search API. Events use the same
github-commit:{sha} ID scheme as github_search for dedup. Per-repo
poller state enables full backfill on first run, page-1-only on
subsequent polls. Weekly poll interval by default.

Closes #1

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

crates/moments-worker/src/main.rs

@@ -7,6 +7,7 @@ use moments_data::{
     bugzilla::{BugzillaConfig, BugzillaSource},
     gitea::{GiteaConfig, GiteaSource},
     github::{GithubConfig, GithubSource},
+    github_repo::{GithubRepoConfig, GithubRepoSource},
     github_search::{GithubSearchConfig, GithubSearchSource},
     hg::{HgConfig, HgSource},
 };
@@ -35,6 +36,11 @@ struct Args {
     #[arg(long, env = "SEARCH_POLL_INTERVAL_SECS", default_value = "86400")]
     search_interval_secs: u64,
 
+    /// Seconds between per-repo commit enumeration polls (full history backfill).
+    /// Defaults to weekly — expensive initial scan, cheap afterwards.
+    #[arg(long, env = "REPO_POLL_INTERVAL_SECS", default_value = "604800")]
+    repo_interval_secs: u64,
+
     #[arg(long, env = "GITEA_HOST", default_value = "git.lair.cafe")]
     gitea_host: String,
 
@@ -132,6 +138,17 @@ async fn main() -> anyhow::Result<()> {
         },
     )) as Arc<dyn EventSource>;
 
+    let github_repo = Arc::new(GithubRepoSource::new(
+        http.clone(),
+        store.clone(),
+        store.clone(),
+        GithubRepoConfig {
+            user: args.github_user.clone(),
+            token: args.github_token.clone(),
+            ..Default::default()
+        },
+    )) as Arc<dyn EventSource>;
+
     let gitea = Arc::new(GiteaSource::new(
         http.clone(),
         store.clone(),
@@ -180,6 +197,7 @@ async fn main() -> anyhow::Result<()> {
         bugzilla_email = args.bugzilla_email,
         events_interval_secs = args.interval_secs,
         search_interval_secs = args.search_interval_secs,
+        repo_interval_secs = args.repo_interval_secs,
         gitea_interval_secs = args.gitea_interval_secs,
         hg_interval_secs = args.hg_interval_secs,
         bugzilla_interval_secs = args.bugzilla_interval_secs,
@@ -188,6 +206,7 @@ async fn main() -> anyhow::Result<()> {
 
     let interval = Duration::from_secs(args.interval_secs);
     let search_interval = Duration::from_secs(args.search_interval_secs);
+    let repo_interval = Duration::from_secs(args.repo_interval_secs);
     let gitea_interval = Duration::from_secs(args.gitea_interval_secs);
     let hg_interval = Duration::from_secs(args.hg_interval_secs);
     let bugzilla_interval = Duration::from_secs(args.bugzilla_interval_secs);
@@ -195,6 +214,8 @@ async fn main() -> anyhow::Result<()> {
     let github_task = tokio::spawn(async move { run_poller(github, interval).await });
     let github_search_task =
         tokio::spawn(async move { run_poller(github_search, search_interval).await });
+    let github_repo_task =
+        tokio::spawn(async move { run_poller(github_repo, repo_interval).await });
     let gitea_task = tokio::spawn(async move { run_poller(gitea, gitea_interval).await });
     let hg_task = tokio::spawn(async move { run_poller(hg, hg_interval).await });
     let bugzilla_task =
@@ -204,6 +225,7 @@ async fn main() -> anyhow::Result<()> {
     info!("shutdown signal received");
     github_task.abort();
     github_search_task.abort();
+    github_repo_task.abort();
     gitea_task.abort();
     hg_task.abort();
     bugzilla_task.abort();