Allow WebSocket to keep sending after close().

In Cloudflare Workers, we've observed applications in the wild that continue to send messages after a Close message. This is incorrect, but it happens, and we end up logging a spurious error as we attempt to proxy the message through.

By removing this restriction, we can now proxy these WebSockets despite the incorrect usage.

Make KJ_CASE_ONEOF work in templated contexts.

Don't escape forward slashes in JSON output

Forward slashes don't need to be escaped in JSON.

Merge pull request #751 from capnproto/harris/zero-length-responses-get-no-content-length-and-great-sadness

Zero-length HTTP responses get no Content-Length header

There is currently no way to explicitly omit a Content-Length/Transfer-Encoding header on an HTTP response to a HEAD request. This is awkward for a proxy, which would ideally pass along responses as-is, even if they have no such headers.

This change allows an author to pass zero as the expected body length to HttpService::Response::send() to mean "do not set any body header." This means that a proxy might strip Content-Length: 0 headers, but will never add a Content-Length header where there was none before.

Silence compiler warning

Implement JSON-RPC adapter to Cap'n Proto interfaces.

This isn't related to JSON-RPC but is a fix needed for the Ekam language server.

This implements JSON-RPC 2.0: https://www.jsonrpc.org/specification

Some protocols, like Visual Studio Code's Language Server Protocol, have made the unfortunate decision to use HTTP-style message envelope even though they are not HTTP protocols. LSP, for example, sends each JSON-RPC messages as a Content-Length header followed by two CRLF followed by a JSON message of that length. I didn't want to rewrite HTTP parsing, so I extended the HTTP library to make this functionality reusable.

Overload Maybe::operator=() to match all copy constructors.

Immovable types can be already wrapped into `Maybe`, which can be constructed as default `nullptr`, set to a constructed value via `.emplace()`, but couldn't be cleared back to `nullptr`. The previous commit fixed the bug, this commit adds a test.

Previously we depended on the compiler to implicitly invoke the copy constructor followed by the assignment operator, but this can lead to the compiler making surprising decisions in some cases, such as deciding that the literal 0 is a better match for nullptr than for T. Overloading operator=() to match every copy constructor is the safe thing to do.

Fix minor typo spotted by lintian

Fix JSON field handlers not triggering on union members of annotated types.

Wrap KJ_LOG in braces

Add a way to set a maximum number of event loop turns before we poll for input.

Implement kj::newTee()

Using "for" statement instead of "if", as suggested by Kenton, still
avoids "dangling else" warnings and also allows to have diagnostics on
missing ";" at the end. So this way is better.

Make it possible to configure HttpClient to not reuse connections at all.

This fixes "dangling else" warnings when KJ_LOG is used after unbraced if statement.

Fix json-test on MSVC.