This change introduces a bug with DELETE request proxying

Hi! This change causes issues when proxying DELETE requests with passThrough: true.

Problem:

When a DELETE request is proxied, this code path now:

1. Deletes content-length header (line ~106 in passThrough logic)
2. Adds transfer-encoding: chunked (because 'delete' is now in CHUNKABLE)
3. But content-length: 0 gets re-added by the underlying HTTP client

Result:

The request ends up with both transfer-encoding: chunked AND content-length: 0 headers, which violates RFC 7230. Most servers (including Express) correctly reject this with 400 Bad Request.

Reproduction:

// Any DELETE request through h2o2 proxy with passThrough: true
h.proxy({ passThrough: true, uri: "http://target-server/endpoint" });

Fix:

DELETE requests typically don't have request bodies, so they shouldn't need chunked encoding. Consider either:

1. Remove 'delete' from CHUNKABLE array, or
2. Add delete options.headers['content-length']; after setting transfer-encoding: chunked

Context:

This breaks serverless-offline HTTP proxy functionality and likely affects other h2o2 users proxying DELETE requests.

Related:

• GitHub Issue #124
• Affects serverless-offline users with HTTP_PROXY configurations