- Add shutdown() method (to be used in forthcoming call combiner code).
- Use a vtable instead of storing method pointers in each instance.
- Check all callers of pull() to make sure that they are properly
  handling errors.
- Clarify ownership rules and attempt to adhere to them.
- Added a new grpc_caching_byte_stream implementation, which is used in
  http_client_filter to avoid having to read the whole send_message byte
  stream before passing control down the stack.  (This class may also be
  used in the retry code I'm working on separately.)
- As part of this, did a major rewrite of http_client_filter, which
  made the code more readable and fixed a number of potential bugs.

Note that some of this code is hard to test right now, due to the fact
that the send_message byte stream is always a slice_buffer stream, for
which next() is always synchronous and no destruction is needed.
However, some future work (specifically, my call combiner work and
Craig's incremental send work) will start leveraging this.

Patch dashboard schema to add new metrics 'Queries/CPU-sec'

Fix issue from google/protobuf#1745 - javascript allow dot in filename

Generate kokoro_index.html before starting kokoro build

Adding user_data to avl interface so that an exec_ctx can be passed in.

gRPC Python test infrastructure.

grpclb: deal with all-drops serverlists

Update package info with classifiers and updated webpage

Disable epoll1 unless explicitly requested

(The time-related first part of it, anyway.)

Revert "Separate internal-only and public parts of C++ API"

Move gtest last

Exit 1 if some tests fail

Remove a follow-on internal that was added post-internalization, keep the templated QPS worker rather than going back to internal API

Fix wrong logging of LB policy addresses

Fix error in Doxygen comment.

Fix memory leak in sent ruby metadata

Support closed fd in grpc_fd_orphan()

Fix ruby test flake

Change security interfaces to better handle both sync and async cases.