Separate KOMODO_STATES and CURRENCIES

[jmjatlanta]

Part of #479

The existing code looks up a pointer to komodo_state in an array, indexed by CURRENCIES.

This PR breaks that link, allowing the collection of komodo_state objects to

• Lazily grow - The komodo_state objects are created as needed
• Logarithmic lookups - the need to look up an index first is not necessary
• hold the symbol - Interrogate the object for the symbol instead of maintaining one locally

This PR also

• hide the CURRENCIES array behind a method. `komodo_currency(uint8_t) will return the currency symbol of a particular index.
• Breaks the two things that happened in komodo_stateptr (getting the symbol and the pointer) into two methods. This allows methods that do not need the symbol and dest values to not create variables for them.
• Override komodo_stateptr to give it the functionality of komodo_stateptrget. This provides a unified interface.

komodo_stateptr() now returns a pointer to komodo_state for the default chain.
komodo_stateptr(char *) now returns a pointer to komodo_state for the passed-in symbol.
komodo_nameset(char*, char*, const char*) can be used to set the symbol and dest as komodo_stateptr used to.

Note: It seems the pax_ routines use the currency array quite a bit. I am unsure of the functionality of this code. This area should be tested heavily to assure no functionality is broken.

[dimxy]

I think these changes are great, however as they touch the notary code I'd suggest someone tries them on a NN node (@DeckerSU)

[dimxy]

probably this PR will need to be rebased after jmj_npoints is merged (f.e. it still uses sp->NUM_NPOINTS replaced on NPOINTS.size() in the jmj_npoints)

[dimxy]

Changes look great and there are tests for them but I still suggest a NN operator would try to use them and check for correct work (absence of errors in logs and console)

[DeckerSU]

I agree with @dimxy , changes looks very good, but before detail review and logic check i should mention one thing, currencies array:

char CURRENCIES[][8] = { "USD", "EUR", "JPY", "GBP", "AUD", "CAD", "CHF", "NZD", // major currencies
    "CNY", "RUB", "MXN", "BRL", "INR", "HKD", "TRY", "ZAR", "PLN", "NOK", "SEK", "DKK", "CZK", "HUF", "ILS", "KRW", "MYR", "PHP", "RON", "SGD", "THB", "BGN", "IDR", "HRK",
    "KMD" };

Is related to PAX "stable-coins" experiment. So, all the code related to these PAX currencies (except KMD of course) is abandoned and will not ever used anymore. My personal feeling of this case - we should carefully remove all PAX related code from daemon, instead of trying to refactor corresponding sources. Probably @ca333 will have other vision, so what about this PR - we should discuss it first, to find a consensus.

My humble opinion and vision, as i said, we should carefully remove all PAX related code from daemon, for this PR it will mean to have only one komodo_state structure related to KMD. Also, in this case we will not need additional class to store all other currencies states. But, once again, this is only my vision and i would like to hear other opinions.

[jmjatlanta]

currencies array:
Is related to PAX "stable-coins" experiment.

If PAX code is indeed dead code, my vote would be to close or delay merging this PR, and scope-creep this into a "Remove PAX" PR.

Update: See PR #551

[TheComputerGenie]

Given that we're giving up on LABS, shouldn't that come out too?

[jmjatlanta]

Closed in favor of #551