Asynchronous detectors and Souffle execution

[jubnzv]

Closes #27

• I have updated CHANGELOG.md
• I have added tests to demonstrate the contribution is correctly implemented
• No test failures were reported when running yarn test-all
• I did not do unrelated and/or undiscussed refactorings

[jubnzv]

It gives a small performance boost comparing with master.

master benchmark results on sample-jetton.tact:

Detector Name	Result
All Detectors	0.49 ops/sec (~2033.16 ms/run) ±9.56%
DivideBeforeMultiply	0.54 ops/sec (~1837.80 ms/run) ±0.98%
ReadOnlyVariables	0.54 ops/sec (~1849.14 ms/run) ±4.56%
NeverAccessedVariables	0.56 ops/sec (~1777.49 ms/run) ±1.55%
UnboundLoops	0.55 ops/sec (~1822.80 ms/run) ±3.80%
ZeroAddress	0.57 ops/sec (~1755.86 ms/run) ±0.74%
ConstantAddress	0.56 ops/sec (~1782.98 ms/run) ±3.75%
BranchDuplicate	0.57 ops/sec (~1757.69 ms/run) ±1.18%
DumpIsUsed	0.56 ops/sec (~1792.60 ms/run) ±4.53%
FieldDoubleInit	0.57 ops/sec (~1759.16 ms/run) ±1.31%
PreferAugmentedAssign	0.57 ops/sec (~1763.16 ms/run) ±4.55%

This PR benchmark results on sample-jetton.tact:

Detector Name	Result
All Detectors	0.54 ops/sec (~1865.07 ms/run) ±6.96%
DivideBeforeMultiply	0.56 ops/sec (~1789.02 ms/run) ±4.39%
ReadOnlyVariables	0.58 ops/sec (~1727.20 ms/run) ±0.66%
NeverAccessedVariables	0.58 ops/sec (~1730.55 ms/run) ±3.65%
UnboundLoops	0.58 ops/sec (~1715.04 ms/run) ±0.85%
ZeroAddress	0.59 ops/sec (~1693.81 ms/run) ±0.71%
ConstantAddress	0.56 ops/sec (~1782.88 ms/run) ±6.80%
BranchDuplicate	0.53 ops/sec (~1901.73 ms/run) ±5.95%
DumpIsUsed	0.56 ops/sec (~1795.69 ms/run) ±1.22%
FieldDoubleInit	0.55 ops/sec (~1813.77 ms/run) ±3.05%
PreferAugmentedAssign	0.57 ops/sec (~1747.10 ms/run) ±1.01%

But the performance difference should be more remarkable when using more complicated Souffle detectors or introducing any parallelizable logic to other detectors (e.g. filesystem operations, API calls, programs execution).