[Repo Assist] perf: bulk memcpy/memset for %s padding and string copy in l_vsnprintf

[github-actions]

🤖 This is an automated pull request from Repo Assist.

Summary

Replace character-by-character loops in l_vsnprintf's %s path with bulk l_memcpy/l_memset calls, giving O(n/word_size) throughput instead of O(n) for padded string printing.

Root Cause

The %s format specifier's padding and copy loops used L_SNPRINTF_EMIT one character at a time, creating per-character overhead for potentially large strings.

Fix

Replace with bounds-safe bulk operations using a L_SNPRINTF_AVAIL() helper macro that prevents size_t underflow when n==0. Truncation and NUL termination semantics are preserved.

Test Status

✅ Linux/gcc: build PASS, 1515 assertions PASS
✅ Linux/clang: build PASS, 1513 assertions PASS
✅ Verify: 52 binaries, max 117KB, static PASS

Generated by 🌈 Repo Assist at {run-started}. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@1f672aef974f4246124860fc532f82fe8a93a57e