Partial fix for issue 303

[pmconrad]

This PR gets rid of the "no blocks to pop" message in witness_node and some unit tests.

IMO a better fix would be to define a clear distinction between "open" and "closed" states in the database, and handle state transitions within the database code. Feel free to reject this PR if you agree. :-)

[oxarbitrage]

tested and works good in the witness node, nice output:

...
1533270ms th_a       application.cpp:538           handle_block         ] Got block: #18670362 time: 2017-07-28T19:25:33 latency: 270 ms from: bhuz  irreversible
: 18670337 (-25)
1534008ms asio       main.cpp:170                  operator()           ] Caught SIGINT attempting to exit cleanly
1534033ms th_a       main.cpp:183                  main                 ] Exiting from signal 2
1534132ms th_a       db_management.cpp:153         close                ] Rewinding from 18670362 to 18670342
root@NC-PH-1346-07:~/bitshares/pull336/bitshares-core# 

However in the tests running a random one as:

./chain_test -t operation_tests/worker_create_test -l all

...
/root/bitshares/pull336/bitshares-core/tests/common/database_fixture.cpp(125): info: check db.get_node_properties().skip_flags == database::skip_nothing passed
2335372ms th_a       db_management.cpp:153         close                ] Rewinding from 2 to 0
2335372ms th_a       db_management.cpp:171         close                ] Database close unexpected exception: {"code":10,"name":"assert_exception","message":"As
sert Exception","stack":[{"context":{"level":"error","file":"fork_database.cpp","line":41,"method":"pop_block","hostname":"","thread_name":"th_a","timestamp":"20
17-07-28T19:38:55"},"format":"_head: no blocks to pop","data":{}},{"context":{"level":"warn","file":"db_block.cpp","line":431,"method":"pop_block","hostname":"",
"thread_name":"th_a","timestamp":"2017-07-28T19:38:55"},"format":"","data":{}}]}
Leaving test case "worker_create_test"; testing time: 33046mks
Leaving test suite "operation_tests"

We could just remove the exception log at https://github.com/pmconrad/graphene/blob/165c914308b8b6df915b9fbd79a3279a1bf2ae2e/libraries/chain/db_management.cpp#L171

It will not be the first catch with no log in the code, there is actually one just above at https://github.com/pmconrad/graphene/blob/165c914308b8b6df915b9fbd79a3279a1bf2ae2e/libraries/chain/db_management.cpp#L164

Output with that change in the same test case will be:

/root/bitshares/pull336/bitshares-core/tests/common/database_fixture.cpp(211): info: check core_in_orders.value == reported_core_in_orders.value passed
/root/bitshares/pull336/bitshares-core/tests/common/database_fixture.cpp(125): info: check db.get_node_properties().skip_flags == database::skip_nothing passed
3275391ms th_a       db_management.cpp:153         close                ] Rewinding from 2 to 0
Leaving test case "worker_create_test"; testing time: 35791mks
Leaving test suite "operation_tests"
Leaving test suite "Master Test Suite"

*** No errors detected
root@NC-PH-1346-07:~/bitshares/pull336/bitshares-core/tests# 

I will approve it as it is with a commit to remove the exception wlog is that is not a problem.

adding @abitmore so he can take another look. good job :)

[pmconrad]

The message still appears in those unit tests that don't create a sufficient number of blocks. In #303 I identified 2 reasons for the message, and this PR fixes only the first.
I'm against removing the log message because the message indicates a problem that should be fixed, see my original comment above.

[abitmore]

Looks good.

[oxarbitrage]

this is good enough, i am merging it.