Add more microbenchmarks to assess interpreter performance

[smarr]

First this PR reorganizes the benchmarks introduced in #119 and renames them.

Benchmarks now try to consistently use a name that roughly describes the expression that is evaluated in the loop body.

Each benchmark follows the same structure that tries to focus the execution of a particular expression. It is executed 30 times inside a loop, which hopefully should make sure that these expressions dominate the overall execution time.

The current micro, or perhaps rather nanobenchmark set is:

Benchmark	Expression	Description
ArgRead	arg	read a method argument
ArrayReadConst	arr at: 1	read a method argument
ArrayWriteConstConst	arr at: 1 put: 2	write a constant to an array at contant index
BlockSend0ConstReturn	[ 0 ] value	activate block that evaluates to constant
Const	42	read a literal
FieldConstWrite	field := 1	write a constant to field
FieldRead	field	read field
FieldReadIncWrite	field := field + 1	read, add 1, and write to a field
FieldReadWrite	field := field	read field and write to field
GlobalRead	GlobalRead	read global
LocalConstWrite	local := 1	write constant to local
LocalRead	local	read local variable
LocalReadIncWrite	local := local + 1	read, add 1, and write to a local
LocalReadWrite	local := local	read and write local
SelfSend0	self trivial	self send message without argument to trivial = ( ^ 1 )
SelfSend0BlockConstNonLocalReturn	self trivial	self send message, which returns nonlocally from an embedded block: trivial = ( [ ^ 42 ] value )

Core-Lib Updates

Status	SOM	PR
✅	SOM (java)	SOM-st/som-java#36
✅	TruffleSOM	SOM-st/TruffleSOM#205
✅	PySOM	SOM-st/PySOM#61
✅	JsSOM	SOM-st/JsSOM#17
✅	SOM-RS	Hirevo/som-rs#46