feat(preconf): harden HTTP server limits

[fridrik01]

This PR hardens the preconf HTTP server and validator client against DoS and OOM, The sequencer is internet-facing, so it previously accepted unbounded request bodies, had no read/write/idle timeouts, and no cap on concurrent connections, leaving it open to OOM, Slowloris, and connection-exhaustion attacks.

More specifically, the main changes are:

• Request body cap (server): MaxBytesReader (2 KB) -> 413 on oversized bodies.
• Read/Write/Idle timeouts (server): 15s / 30s / 60s, closes Slowloris and idle keep-alive vectors.
• Max concurrent connections (server): netutil.LimitListener(256).
• Response body cap (client): default 16 MiB, configurable via max-response-size, to prevent a malicious sequencer OOMing a validator. Picked 16 MiB which should cover the worst-case valid block, full hex-encoded transactions (~6 MiB) + 6 blobs (~1.5 MiB) with ~2× headroom.

[codecov]

Codecov Report

❌ Patch coverage is 86.66667% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 60.22%. Comparing base (e23dc05) to head (8684280).

Files with missing lines	Patch %	Lines
beacon/preconf/config.go	25.00%	2 Missing and 1 partial ⚠️
node-core/components/preconf_client.go	0.00%	1 Missing ⚠️

Additional details and impacted files

@@                       Coverage Diff                        @@
##           tls-preconf-server-hardening    #3119      +/-   ##
================================================================
+ Coverage                         60.01%   60.22%   +0.21%     
================================================================
  Files                               384      384              
  Lines                             20055    20079      +24     
================================================================
+ Hits                              12036    12093      +57     
+ Misses                             7028     6986      -42     
- Partials                            991     1000       +9     

Files with missing lines	Coverage Δ
beacon/preconf/client.go	52.20% <100.00%> (+30.33%)	⬆️
beacon/preconf/server.go	77.20% <100.00%> (+1.88%)	⬆️
beacon/preconf/server_metrics.go	100.00% <ø> (ø)
node-core/components/preconf_client.go	16.66% <0.00%> (-0.36%)	⬇️
beacon/preconf/config.go	81.25% <25.00%> (-8.04%)	⬇️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[calbera]

lgtm but can we add e2e tests for the scenarios you mentioned were previously possible (but shouldn't be anymore)?

previously accepted unbounded request bodies, had no read/write/idle timeouts, and no cap on concurrent connections

[bar-bera]

agree on testing. Also MaxHeaderBytes (DefaultMaxHeaderBytes = 1 << 20 // 1 MB) and MaxResponseHeaderBytes (default at 10 MB) are not set so the the effective per-request memory may be underestimated, not game changing since connections are already limited but maybe you want to consider setting them.