Fix memory leak (24GB per second) when rodeos starts with --filter-wa…

[linhuang-blockone]

…sm and is not connected to nodeos

Change Description

When rodeos' ship_client fails to connect to nodeos (like when nodeos is not running),. ship_client calls cloner_session's close() with retry set to true. This triggers cloner_plugin to recreate cloner_session every second until a connection is established. cloner_session creates rodeos_filter object if rodeos starts with --filter-wasm option. Although destructor of rodeos_filter is called before recreation, its memory is not completely freed. The root cause is the filter_state in rodeos_filter has a member eosio::vm::wasm_allocator wa. wa allocates VM memory when initiated. This memory must be released by explicit free().

The same problem in thread_state is fixed by this PR as well.

Change Type

Select ONE:

• Documentation

• Stability bug fix

• Other

• Other - special case

Testing Changes

Select ANY that apply:

• New Tests

• Existing Tests

• Test Framework

• CI System

• Other

Consensus Changes

• Consensus Changes

API Changes

• API Changes

Documentation Additions

• Documentation Additions