Fix LightGBM models locale sensitivity and improve R/W performance.

[AlbertoEAF]

When Java is used, the default C++ locale is broken. This is true for
Java providers that use the C API or even Python models that require JEP.

This patch solves that issue making the model reads/writes insensitive
to such settings.
To achieve it, within the model read/write codebase:

• C++ streams are imbued with the classic locale
• Calls to functions that are dependent on the locale are replaced
• The default locale is not changed!

This approach means:

• The user's locale is never tampered with, avoiding issues such as
  [CRITICAL BUG][Python] cannot wrire() UTF-8 strings by UnicodeEncodeError microsoft/LightGBM#2979 with the previous
  approach Fix Booster read/write locale dependency microsoft/LightGBM#2891
• Datasets can still be read according the user's locale
• The model file has a single format independent of locale

Changes:

• Add CommonC namespace which provides faster locale-independent versions of Common's methods
• Model code makes conversions through CommonC
• Cleanup unused Common methods
• Performance improvements. Use fast libraries for locale-agnostic conversion:
  • value->string: https://github.com/fmtlib/fmt
  • string->double: https://github.com/lemire/fast_double_parser (10x
    faster double parsing according to their benchmark)

Bugfixes:

• Booster.predict returns same score for any distinct instance when executed via JEP microsoft/LightGBM#2500
• [bug] SWIG Booster read/write dependent on machine locale microsoft/LightGBM#2890
• Invoking LightGBM via JEP has distinct behaviour than invoking outside of JEP (simple python script) ninia/jep#205 (as it is related to LGBM as well)

[AlbertoEAF]

This is our internal PR to fix the model locale. The PR to the Microsoft codebase is independent and lives at microsoft#3405.

[shengwangsw]

To simplify, I looked at the code, basically, you added submodules fmt and fast double parser with its functionalities as you described at the description. And then you change the code that every line that uses common to commonC parser that you created. I believe that it was already discussed and I didn't see anything wrong. Yet try to understand why did travis build fail

[paulojrp]

good job :)

[AlbertoEAF]

@shenggwang can you approve?
Ignore the checks. The java build works which is the one we're interested in.

[shengwangsw]

LGTM

[AlbertoEAF]

As discussed, this branch will still receive more commits to make it acceptable for Microsoft's mainline code.

As such, a tag v3.0.0-with_model_locale_fix_for_java was added at this point and will be used in the provider (feedzai/feedzai-openml-java#53).