Added the HTTP server and serve command, with proxy, CSRF, TLS and security header config; wsserve is now ws serve, alongside a new ws dial

The serve command runs internal/server: the API under /api (health, and
the CSRF token endpoint) and the WebSocket hub under /ws, over HTTP or TLS,
with an optional HTTP to HTTPS redirect, the PROXY protocol from trusted
sources and a graceful shutdown that closes WebSockets first. Its middleware
covers the client IP behind trusted proxies, request IDs, request logging,
panic recovery, global and API rate limits, Host and body size checks,
security response headers, compression, CSRF protection (cross-origin
checks and signed double-submit tokens) and JSON-only API bodies. Middleware
that logs takes a WithLogger option.

New server config sections: proxy, csrf, tls, headers and frontend (not
used yet), plus max_header_bytes, max_body_bytes, global and API rate limits
and HTTP read, write and idle timeouts. Every setting documents what 0 or
empty means, and internal/config/README.md is reorganised by section.
server.timeouts.read_header must be greater than 0, and a header or idle
timeout of 0 means no limit rather than the read timeout.

WebSocket upgrades are protected by the Origin check: the server's own
origin, or one of csrf.cross_origin_protection.trusted_origins, which the
hub now takes with WithTrustedOrigins.

wsserve becomes "ws serve". "ws dial" opens a WebSocket on /ws to test it:
--ping sends a ping and waits for the pong, --message sends messages and
prints the replies, --origin tests the Origin check, and with neither
--ping nor --message each line of stdin is sent until Ctrl+C.

CODE_CONVENTIONS.md

@@ -754,14 +754,20 @@ says what 0 or empty means, and the code treats it consistently:
 
 - A timeout of 0 or less means **no limit**, and an interval of 0 or less
   means **off** (e.g. `ping_interval`), never "already expired". Where off
-  makes no sense the setting rejects 0 instead (a source's `interval`: a
-  source is turned off with `enabled`). Code applies a timeout with a helper
-  rather than `context.WithTimeout(ctx, 0)`:
+  makes no sense, or no limit is unsafe, the setting rejects 0 instead (a
+  source's `interval`: a source is turned off with `enabled`;
+  `server.timeouts.read_header`: a client could hold a connection open for
+  ever). Code applies a timeout with a helper rather than
+  `context.WithTimeout(ctx, 0)`:
 
   ```go
   writeCtx, cancel := util.WithTimeout(ctx, conf.WriteTimeout) // no deadline if 0 or less
   ```
 
+  Where a library reads 0 as some other timeout (`http.Server`'s
+  `IdleTimeout` uses `ReadTimeout`), the code passes it a negative timeout
+  instead, which such libraries read as no limit.
+
 - A size of 0 means **the default** (`max_message_size` of `0B` reads
   32KiB), and -1 means **no limit** (`util.ByteSizeNoLimit`); no other size
   may be negative. A setting for which no limit is unsafe rejects -1 in its