SDL Compliance Completion

[agniuks]

Adds two forked folly files in order to re-enable C4244 and C4267 warning suppressions. I opened a PR in folly to address this from there, and will add an issue to track updating folly and removing these files.
Closes #6918.

Microsoft Reviewers: Open in CodeFlow

[asklar]

Left some questions.

Meta point (this one shouldn't block this PR), it might be better to use patches instead of forked files to be able to PR the individual forked change rather than an entire file that has maybe one line different. Thoughts?

[agniuks]

Left some questions.

Meta point (this one shouldn't block this PR), it might be better to use patches instead of forked files to be able to PR the individual forked change rather than an entire file that has maybe one line different. Thoughts?

@asklar That sounds like the better approach, especially with the minimal changes. I can update this to use patches instead.

edit: Will push this to get the changes in now and will follow up on potentially changing this to patch files separately.

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[asklar]

[talked offline, we still need to figure out how x86 should/will work]

[agniuks]

@asklar, I updated to_ascii_port_clzll() method to call builtin_clzll() on x86. builtin_clzll() will then call __builtin_clz() which is supported only in VS 16.3-16.8 but code added in folly handles the logic for calling polyfill that uses _BitScanReverse when __builtin_clz is not available

With the forked files it's hard to see the diffs; for reference the only changes I've made to these files from what we were using before are in ToAscii.h lines 59-63(adding check for x86), 64/69(updates to match changes upstream - return type changed to auto from int and added return statement in line 69), 84(added cast), & 233(added cast). in Dynamic.cpp the only change is an added cast on line 300.