Add compile-time option TRACK_STANDBY_VOTES #987

[abitmore]

PR for #987.

[jmjatlanta]

Just to make sure I am understanding the issue correctly:

This will help the API nodes get correct tallies of votes for inactive objects without slowing down other nodes that don't care about vote counts for inactive objects.

The code looks great. I plan to take a deeper dive soon.

Tests would be nice, but testing compile switches is a pain and near impossible to do in an automated way.

Just a note here so we don't forget: Documentation of this switch should be made available.

[jmjatlanta]

Tested with Ubuntu 18.04 (Boost 1.65/OpenSSL 1.1) with option STANDBY_VOTES_TRACKING ON as well as OFF. Everything worked as expected.

A unit test would be nice, even if it only checked the expected results depending on the currently compiled option.

[oxarbitrage]

@jmjatlanta we need to merge this one before: #1099
#1099 one haves a test case: #1169

we should do the same here.
do you have time to make it ?

the flow to merge the last @abitmore pull requests can be:

• merge Improve adjust_balance performance #1083 #1099
• merge Add compile-time option TRACK_STANDBY_VOTES #987 #1140
• merge add asset api test case #1169
• merge test cases for 1140 if we do them.
• merge add USE_PROFILER to Cmake #1119 -> need cmake outout msg that build will be made with profile.

[jmjatlanta]

The test case in #1169 is a great template for this one (and 1099 as well).

I will create the tests in different but similarly named files to avoid merge conflicts. We can merge them into 1 file later.

I will probably need help verifying that I'm creating the environment that tests the actual change.

[oxarbitrage]

as discussed in telegram; @jmjatlanta will review #1099 and #1169 and we can merge them those if they are ready.

@oxarbitrage will do the test case for #1140

[pmconrad]

Tests would be nice, but testing compile switches is a pain and near impossible to do in an automated way.

Good point. How about making it a runtime config switch instead. We already have a configurable setting in database, namely the checkpoints list, so I suppose adding more options is not an architectural breach. Opinions?

[abitmore]

Good point. How about making it a runtime config switch instead. We already have a configurable setting in database, namely the checkpoints list, so I suppose adding more options is not an architectural breach. Opinions?

Will do. Actually run-time option is required in the user story.

Will update code

[abitmore]

Replaced by #1191