Implement Integer.sqrt for large integers

[tompng]

Implement Integer.sqrt algorithm used in CRuby (Newton's method)
ruby/ruby#10274

This spec now passes

Integer.sqrt(10**400).should == 10**200

Calculation

(x + n / x) / 2.0 is always greater or equal to square_root(n) if there is no calculation error.
Integer arithmetic (x + n / x) / 2 is greater than square_root(n) OR equal to square_root(n).floor
After x -= 1 while x*x > n, x will be the expected return value of Integer.sqrt(n)
Newton's method ensures this while loop is executed only a few times. Experimentally, it's executed only once.

[oracle-contributor-agreement]

Thank you for your pull request and welcome to our community! To contribute, please sign the Oracle Contributor Agreement (OCA).
The following contributors of this PR have not signed the OCA:

• PR author: tompng
• tomoyapenguin@gmail.com (@tompng)

To sign the OCA, please create an Oracle account and sign the OCA in Oracle's Contributor Agreement Application.

When signing the OCA, please provide your GitHub username. After signing the OCA and getting an OCA approval from Oracle, this PR will be automatically updated.

If you are an Oracle employee, please make sure that you are a member of the main Oracle GitHub organization, and your membership in this organization is public.

[oracle-contributor-agreement]

Thank you for signing the OCA.

[eregon]

Very nice, thank you for porting this optimization and correctness fix to TruffleRuby!
Ruby code looks a lot cleaner for this compared to C code :)

[eregon]

@andrykonchin Could you integrate this and look if there are some CRuby tests for this and we could remove excludes for them?
Also please add a CHANGELOG entry.

[andrykonchin]

Wondering how the constant 0xfffffffffffff was chosen. Is it based on benchmarking results?

[tompng]

Larger is better as long as Math.sqrt(n).floor returns correct value.
0xfffffffffffff is 52bit integer and double has 53bit precision.

If n is larger than 0xfffffffffffff == 0x4000000**2-1, Math.sqrt(n).floor might not return correct value.
Integer.sqrt(0x4000001**2-1) should return 0x4000001 - 1
Math.sqrt(0x4000001**2-1).floor is 0x4000001