perf(aws-lambda): optimize header parsing to reduce micro-allocations

[fcarvajalbrown]

Description

This PR optimizes the getHeaders method in the AWS Lambda adapter. By switching from Object.entries() to a for...in loop with a safety check, we avoid the overhead of creating intermediate arrays for every request.
In high-throughput serverless environments, reducing these "micro-allocations" significantly lowers Garbage Collection (GC) pressure and minimizes latency jitter.

Performance Details

benchmark avg (min … max) p75 / p99

---

Current (Object.entries) 1.26 µs/iter 1.20 µs / 2.30 µs
Optimized (for...in) 851.85 ns/iter 860.13 ns / 954.57 ns

Summary:
Optimized (for...in) is 1.48x faster than Current (Object.entries)

Testing

[x] Benchmark Validation: Verified with mitata (results above).

[x] Functional Testing: Ran the full Hono test suite.

[x] Runtime Verification: Verified that globalThis.Headers is correctly populated.

[x] Logic Safety: Added hasOwnProperty check to ensure we only process own properties of the event.headers object.

[usualoma]

Hi @fcarvajalbrown
Thank you for creating the pull request.

For future pull requests, please include a description following this template:
https://github.com/honojs/hono/blob/main/.github/pull_request_template.md

Regarding this pull request, I believe it is a reasonable small improvement.