Fix update_version.py, so that PROTOBUF_VERSION is always a 9 digit number

Update protobuf version for Java and PHP

Cherry Pick (#6614)

* Sync upb

https://github.com/protocolbuffers/upb/pull/209

* Update upb

https://github.com/protocolbuffers/upb/pull/210

Java: Add Automatic-Module-Name entries to the Manifest

This commit removes the question template. It also updates the bug report and feature request templates to direct questions to the mailing list.

* When length is zero, substr returns null instead of emptry string, which breaks the invariable for message.
* Tested in https://github.com/protocolbuffers/protobuf/pull/6560

Update version number

* For missing message value, map should create a default message
instance in order to keep its invariable.
* On 32-bit platform, int64 map key should be string

We've notice significant issues downloading from zlib.net. Since there
are already other archives coming from GitHub, this shouldn't negatively
affect reliability.

Optimization for layout_init()

* Add missing files under benchmarks/ to the dist

* fix typo

* Sync upb

* https://github.com/protocolbuffers/upb/pull/208

* Fix php c extension compiling

Don't use macros defined by upb

* Update ruby conformance failure list

Enums use varint, so if the value is negative it should end up being longer.

This was caught my new conformance test cases.

Update versions to 3.9.1

Removed ByteCountInt64.  This is a transitional typedef that should n…

* Test singular fields are encoded in canonical way

* Defautl values in proto3 should not be encoded.
* Values should be converted to the canonical representation (e.g.,
long int64 value may be truncated for int32 field)

* Update failure list

* Update failure list

clang-cl requires the order to be the opposite of what it currently is.
It should not affect other compilers since this is the only compiler
that has both macros defined.

Down integrate internal changes to GitHub

Update to Truth 1.0

The previous issue of mktime doesn't work on Mac has been fixed.

We were creating a map decoding frame when starting the *map*,
but clearing the GC slot when finishing each *map entry*.  This
means that the decoding frame could be collected in the meantime.