feat(gnovm)!: store refactor

[thehowl]

This draft PR introduces a re-organized Store interface, with a few goals as compared to the current Store implementation:

• Split the Store interface into smaller interfaces, and replace usages of the Store interface in the GnoVM codebase to only use the subset they need.
  • This has the main advantage of simplifying unit testing and mocking; allowing to create smaller implementations for what needs to be tested.
  • But furthermore, it makes explicit the extent to which a given function will attempt to access the Store.
• Split the functionality of existing stores into wrappers with the philosophy of "do one thing and do it well", and allow composition of different stores depending on necessity.
  • As an example, an ObjectStore will eventually be 3 store implementations stacked together:
    1. A "reference manager" store, which when marshaling changes "compacts" many values to RefValues, and many types to RefTypes; and when unmarshaling expands RefTypes to their corresponding types. (This is done currently by functions like copyValueWithRefs / fillTypesOfValue).
    2. A "cache" store, which holds the result of unmarshaling values from the Store, and of new value
    3. The underlying ObjectStore, which retrieves and sets data in the tendermint 2 store.
  • Another type of store will be the "transaction log" store, which would exist for the current cacheTypes and cacheNodes: it keeps the values to be written, until Write() is called and the values are committed to their underlying "cache" maps.
• Remove any methods which could be functions or options; or other specific implementations
  • For instance, now that PackageGetter is never used on-chain, the Store itself does not need to have a PackageGetter; the TestStore in gnovm/tests can make its own implementation which does the on-the-fly loading of packages and standard libraries.
  • In another instance, instead of setting NativeResolver and PackageInjector we can now set them at startup using StoreOptions - seeing as in our practical usages these are initialised once then never again
• Ensure that the Store interfaces can be better adapted for implementations in non-blockchain contexts, rather than trying to fit defaultStore for all use-cases.

More development and notes will follow. For now, let's try to merge #2319 as a stepping stone first.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 54.88%. Comparing base (74df3ea) to head (d6b813d).

Additional details and impacted files

@@                    Coverage Diff                    @@
##           dev/morgan/issue-2283    #2655      +/-   ##
=========================================================
- Coverage                  54.89%   54.88%   -0.01%     
=========================================================
  Files                        593      593              
  Lines                      81579    81567      -12     
=========================================================
- Hits                       44782    44770      -12     
+ Misses                     33382    33380       -2     
- Partials                    3415     3417       +2     

Flag	Coverage Δ
contribs/gnofaucet	14.46% <ø> (ø)
contribs/gnomd	0.00% <ø> (ø)
misc/autocounterd	0.00% <ø> (ø)
misc/genstd	80.54% <ø> (ø)
misc/goscan	0.00% <ø> (ø)
misc/logos	17.38% <ø> (-0.31%)	⬇️
misc/loop	0.00% <ø> (ø)
tm2	54.46% <ø> (-0.04%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.