Weird observations in VariableEncodingBenchmark

[Jo Durchholz]

Hi all,

just looking at things and noticing unexpected stuff, not sure which of
it is normal and which isn't.
I only looked at VariableEncodingBenchmark because it was giving me
NullPointerExceptions, but then noticed more in the static inner
classes, so here goes:

1) The classes are all named ReadSomething, but they all do new
VariableEncodingBenchmark.writeSomething(this).
1a) With the exception of ReadVarInt, which actually does
VariableEncodingBenchmark.readVarInt(this).

2) IntelliJ tells me that the ReadVarLong class is not referenced from
anywhere.
That's often fine because IntelliJ can miss referenced generated from
reflective code, but not in this case: ReadInt, ReadVarInt, and ReadLong
are referenced from JMH-generated code.
Double-checking by searching the generated sources and with breakpoints
in a nonforked run confirmed that ReadVarLong is, indeed, never
referenced resp. instantiated.

3) The classes that are called always fails in their constructors with a
NullPointerException in the VariableEncodingBenchmark() constructor
because state is still null.

This all isn't going to affect what I'm planning to do, so it's outside
of my current scope of interest, but I thought the dev team might want
to know about these things.

Regards,
Jo