feat(node/http): minor HTTP Server/Response improvements

[talentlessguy]

This PR adds a few improvements (sorry that multiple changes in one PR, I was just fixing a lot of stuff at the same time)

ref: https://nodejs.org/api/http.html

Server

• extends from EventEmitter and has close and connection events
• has a close() method which triggers listener.close()

ServerResponse

• has statusCode and statusMessage properties
• uses web Headers API instead of a string to string record for headers management
• added removeHeader, getHeaderNames, hasHeader
• ensureHeaders is a private method now because it's not defined in Node.js HTTP docs
• setHeader now returns response object for chaining (as defined in Node.js HTTP docs)

[CLAassistant]

All committers have signed the CLA.

[AaronO]

Some overlap with #1435, would be best to merge both improvements.

[bartlomieju]

@talentlessguy could you please combine changes from #1435 in your PR? I would also be great to have some tests covering the changes, especially for all those header methods.

[AaronO]

@talentlessguy Do you mind if I push http changes from #1435 to this branch ? (or maybe it's best to just land this and cherrypick the #1435 changes in another PR)

Note that class field initializers absolutely murder performance, given the importance of the HTTP-stack hot paths, we're quite sensitive to this. An improvement has been landed in v8 (https://chromium.googlesource.com/v8/v8/+/713ebae3b4ef42d00220097ab2f238ffe8e4b87e), I haven't tested it yet, I don't think it closes the entire gap, ideally we should fix this upstream in v8 instead of working around it.

[talentlessguy]

@AaronO yeah it's completely fine

[AaronO]

LGTM

Note: field initializers are atrociously slow, but there are improvements coming in v8 9.7, so we'll punt on them until testing them