Share Name.Literal strings

[JaroslavTulach]

Pull Request Description

Fixes #10330 by interning Name.Literal strings.

Checklist

Please ensure that the following checklist has been satisfied before submitting the PR:

• All code follows the
  Scala,
  Java,
• Unit tests have been written where possible.
• Benchmarks are OK

[Akirathan]

So what is the difference of memory consumption by all the Name.Literal objects on the heap? Is there any difference in sizes of caches?

[JaroslavTulach]

Previously there was almost 6000 of strings representing Standard in memory when running test/Base_Tests:

now, after the fix we have just one:

The overall amount of memory occupied by java.lang.String instances went down by 5MB.

[JaroslavTulach]

Is there any difference in sizes of caches?

I don't see any difference in the size of caches. Probably because of

enso/lib/java/persistance/src/main/java/org/enso/persist/PerGenerator.java

Line 102 in 269354e

obj = s.intern();

the intern() being already called before storing the strings.

[JaroslavTulach]

Of course, calling String.intern() slows things down. Hopefully the showdown isn't going to affect our benchmarks:

• engine benchmarks run
• stdlibs benchmarks run

Benchmarks don't seem to be affected by this change, but while investigating them I realized #11901. Anyway with 13db21c test we are ready to integrate.