Implement Binary codec

[cedy]

Which problem is this PR solving?

Resolves #224

Short description of the changes

Implements extract and inject method of the Binary codec to be able to
inject/extract span context to/from bytearray.

This commit implements binary context injection/extraction the same way as in jaeger-client for go. Inject method dumps span context into bytearray carrier and packs it in big endian order as follow:
8 bytes uint64 for first 8 bytes of trace id, 8 bytes uint64 for last 8 bytes of trace id, 8 bytes uint64 for span id, 8 bytes uint64 for parent id, 1 byte uint8 for flags, 4 bytes unit32 for number of baggage items.
If baggage exists, its packs it as follows:
4 bytes unit32 for number of bytes in key-name, key-name bytes, 4 bytes unit32 for number of bytes in value, value-bytes.
Extract method follows the same pattern.

Binary codec is compatible with golang jaeger-client implementation.

[yurishkuro]

please make sure make lint is successful

[cedy]

please make sure make lint is successful

Lint issues are now resolved

[codecov]

Codecov Report

Merging #304 (6d4b0e3) into master (1cb8e52) will decrease coverage by 0.00%.
The diff coverage is 95.55%.

@@            Coverage Diff             @@
##           master     #304      +/-   ##
==========================================
- Coverage   95.40%   95.40%   -0.01%     
==========================================
  Files          25       25              
  Lines        1937     1980      +43     
  Branches      266      273       +7     
==========================================
+ Hits         1848     1889      +41     
  Misses         56       56              
- Partials       33       35       +2     

Impacted Files	Coverage Δ
jaeger_client/codecs.py	97.02% <95.55%> (-0.38%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1cb8e52...6d4b0e3. Read the comment docs.

[yurishkuro]

Weird, one of the tests failed, PTAL. I would've thought this change would be mostly language-version agnostic.

Unit Tests / unit-tests (py 3.5 tornado >=4,<5) (pull_request)

____ TestCodecs.test_binary_codec_extract_compatibility_with_golang_client _____
tests/test_codecs.py:479: in test_binary_codec_extract_compatibility_with_golang_client
    assert carrier == bytearray(span_context_serialized)
E   AssertionError: assert bytearray(b'a...0\tvalue_one') == bytearray(b'a]...0\tvalue_two')
E     At index 40 diff: 9 != 7
E     Use -v to get the full diff

[cedy]

Weird, one of the tests failed, PTAL. I would've thought this change would be mostly language-version agnostic.

Unit Tests / unit-tests (py 3.5 tornado >=4,<5) (pull_request)

____ TestCodecs.test_binary_codec_extract_compatibility_with_golang_client _____
tests/test_codecs.py:479: in test_binary_codec_extract_compatibility_with_golang_client
    assert carrier == bytearray(span_context_serialized)
E   AssertionError: assert bytearray(b'a...0\tvalue_one') == bytearray(b'a]...0\tvalue_two')
E     At index 40 diff: 9 != 7
E     Use -v to get the full diff

The problem is that python has deterministic dictionary order only from 3.7 version https://stackoverflow.com/a/14959001 . So, if we have more than 1 item in baggage, we can't determine how they are going to be dumped into byte array, therefore byte arrays don't align. I don't see how to find the way around it, but only to have 1 item in the baggage.

[yurishkuro]

Fair enough. Another failure is with Python 2.7:

_________________________ TestCodecs.test_binary_codec _________________________
tests/test_codecs.py:405: in test_binary_codec
    extracted_span_context = tracer.extract(Format.BINARY, carrier)
jaeger_client/tracer.py:273: in extract
    return codec.extract(carrier)
jaeger_client/codecs.py:182: in extract
    key, value, bytes_read = self._unpack_baggage_item(baggage_data)
jaeger_client/codecs.py:206: in _unpack_baggage_item
    value, b_read = self._read_kv(baggage[bytes_read:])
jaeger_client/codecs.py:214: in _read_kv
    return b''.join(data_value).decode('utf-8'), bytes_read
/opt/hostedtoolcache/Python/2.7.18/x64/lib/python2.7/encodings/utf_8.py:16: in decode
    return codecs.utf_8_decode(input, errors, True)
E   UnicodeDecodeError: 'utf8' codec can't decode byte 0xff in position 3: invalid start byte

I am not opposed to dropping 2.7 from CI since it's past EOL since 2020-01-01, but we'd need a different PR to remove it first before merging this one.

[yurishkuro]

#306