move confirm block count to the start_block call, make tester track its own watermarks the same way any producer plugin would, fix off-by-one caused by not accounting for a block confirming itself; fix off-by-one caused by propagating the value of dpos-lib before it was updated; decoupled dpos-lib from bft-lib though controller still returns the max of them

move the deferred limit to the producer plugin which now controls it; route all new blocks through producer so it can maintain invariant that there is always a pending block (whether for speculative execution or future signing); re-apply unapplied transactions and apply deferred transactions on a new block (NOTE: this needs proper throttling once subjective failures are handled correctly inside controller).  Remove spurious calls to start_block in other plugins

remove explicit calls to abort/start_block in favor of asserting that there is/is-not a pending block when it is expected

cleanup of transaction_context, throwing soft resource limit exceptions if limit is due to block, and preferential network billing of prunable data (#2000)

- udpate system contract to set initial account bw to 0
- update system contract to process new on block header
- update cleos api and get info api

- this change prevents plugins from causing a node to reject valid
blocks.

* clean up configuration parameters
* use dynamic max_inline_action_size configuration parameter to limit size of inline actions
* use dynamic max_inline_action_depth to limit recursion depth of inline actions
* change deferred_trx_expiration_window parameter into a dynamic configuration parameter
* add numerator and denominator configuration parameters for later context free data discount implementation
* bill delayed transaction for the additional net usage of the eventual retirement of the deferred transaction

* cpu_usage limit for an action is properly tracked even through inline actions
* receipt should store net usage in words
* total billed cpu usage for a transaction should be rounded up to nearest multiple of 1024
* enforce the cpu and net usage limits set in chain configuration
* additional stateless validation on generated transactions
* clear expired input transactions (and do not touch generated transactions) on start block rather than finalize block
* fix clear expired input transactions to use pending block time rather than head block time
* update producer authority at start of block (implemention still necessary)
* move resource limit updates and updating tapos from commit_block to finalize_block
* fix to controller::last_irreversible_block_id()
* reduce default_base_per_transaction_net_usage

Also fixed special accounts initialization and special_accounts_tests.

Moved special_accounts_tests and wallet_tests into unittests.

multisig_tests fixed, transaction_receipt::scheduled set and used, transaction_metadata::on_result changed to just std::function

Since the delay has now moved into the authority, the delay tests were not setting the delay properly in the updateauth actions.
This meant that a delay was not enforced and the tests were not actually testing delays properly.

I fixed the tests to properly set the delays, but this means that additional tests in delay_tests now fail.
The additional tests that failed are: link_delay_permission_change_test, link_delay_permission_change_with_delay_heirarchy_test, link_delay_link_change_test, and link_delay_link_change_heirarchy_test.
The reason these fail is very likely due to off-by-one errors due to change of now() from head block time to pending block time.

In this commit, I fixed link_delay_permission_change_test but have not yet gotten to the other three tests.

Also, of the two delay tests that were originally failing, I fixed canceldelay_test but have not yet fixed canceldelay_test2.

- block log now deals in signed_block_ptr rather than
optional<signed_block>
- replay / resync works
- logging to block log as commit or push block

Record processed non-implicit transactions into unapplied_transactions during abort_block.

Properly set signed_id in transaction_metadata.

Fix block production in tester (also added ability to make empty blocks).

Also fixed bug in fork_database::prune that could cause memory access violations.

fork_database::get_block_in_current_chain_by_num should returns null if not found instead of asserting.

Also, it now checks to ensure the proposed producer schedule would actually change the schedule.