feat: dynamic health test concurrency

Auto-scale test concurrency to ~10% of proxy count, capped by
test_concurrency config ceiling (default raised from 5 to 25).
Prevents saturating upstream Tor when pool size varies.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

docs/USAGE.md

@@ -77,7 +77,7 @@ proxy_pool:
     - www.cloudflare.com
     - www.amazon.com
   test_timeout: 15
-  test_concurrency: 5
+  test_concurrency: 25       # max parallel tests (auto-scales to ~10% of pool)
   max_fails: 3
   state_file: ""      # empty = ~/.cache/s5p/pool.json
 ```
@@ -180,7 +180,7 @@ proxy_pool:
     - www.cloudflare.com
     - www.amazon.com
   test_timeout: 15          # per-test timeout (seconds)
-  test_concurrency: 5       # parallel health tests
+  test_concurrency: 25      # max parallel tests (auto-scales to ~10% of pool)
   max_fails: 3              # evict after N consecutive failures
   state_file: ""            # empty = ~/.cache/s5p/pool.json
   report_url: ""            # POST dead proxies here (optional)
@@ -209,6 +209,10 @@ by performing a TLS handshake against one of the `test_targets` (rotated
 round-robin). A successful handshake marks the proxy alive. After `max_fails`
 consecutive failures, a proxy is evicted.
 
+Concurrency auto-scales to ~10% of the proxy count, capped by
+`test_concurrency` (default 25, minimum 3). For example, a pool of 73 proxies
+tests 7 at a time rather than saturating the upstream Tor node.
+
 Before each health test cycle, the static chain is tested without any pool
 proxy. If the chain itself is unreachable (e.g., Tor is down), proxy tests
 are skipped entirely and a warning is logged. This prevents false mass-failure