Adding extension points for security context.

Avoid static initialization of the kIdentity block

Updated protobuf to fix gmock url in autogen.sh

slight rewording of grpc_error ownership rules

Specify the "g" in gRPC

Add ServerReaderWriterInterface

Fix fd_orphan in ev_epoll_linux.c

Fix a typo in include/grpc++/impl/codegen/client_context.h

Delete unused TIMEOUT enum in end2end tests

Rename grpc_client_config to grpc_resolver_result.

PHP: reject metadata keys that are not legal

Mark epoll_{,p}wait as blocking regions if they have a timeout

remove skipIf from TypeSmokeTest (issue 7672)

Remove the extra comma

spec: Percent-encode Status-Message

Bazel: export LICENSE file.

merge

Unlike other Objective-C objects, there's no hard reason why the compiler wouldn't be able to initialize a block statically (as it does with NSString literals). And it certainly doesn't complain about it (like it does with other object initializers). But as I haven't been able to find confirmation of this, and we're seeing a weird crash occur near this code, let's play it safe.

Signed-off-by: Piotr Sikora <piotrsikora@google.com>

Clean up cq_verifier code and error messages.

Upmerge from v1.0.x