Task errors and agent message logs

[nerfZael]

Changes:

• Added table task_errors since a task can have multiple errors (e.g. when running, when building transactions, etc). Tasks still has an error column, but that is now only used for errors that stop AutoTx from running
• Increased max consecutive auto reply to 50 in main chat and 20 in delegate researcher (was 10 and 10) since this is effectively the same as max_rounds for our case
• Implemented logging for all agent messages while auto tx is running. This should help greatly when debugging.
• Endpoints for fetching logs

[nerfZael]

/workflows/benchmarks agents/token/send 3

[github-actions]

Finished benchmarks
Download artifacts

Test Run Summary

• Run from: ./autotx/tests/agents/token/send
• Base path: autotx/tests/agents/token/send/test_send.py::
• Iterations: 3
• Total Cost: $0.10
• Total Success Rate (%): ${\color{none} \LARGE \texttt {100.00} \large \texttt {} }$

Detailed Results

Test Name	Success Rate (%)	Passes	Fails	Avg Time	Avg Cost
test_send_erc20	${\color{none} \large \texttt {100} \normalsize \texttt {} }$	${\color{none} \large \texttt {3}}$	${\color{none} \large \texttt {0}}$	21s	$0.00
test_send_erc20_parallel	${\color{none} \large \texttt {100} \normalsize \texttt {} }$	${\color{none} \large \texttt {3}}$	${\color{none} \large \texttt {0}}$	23s	$0.01
test_send_eth_multiple	${\color{none} \large \texttt {100} \normalsize \texttt {} }$	${\color{none} \large \texttt {3}}$	${\color{none} \large \texttt {0}}$	28s	$0.01
test_send_native	${\color{none} \large \texttt {100} \normalsize \texttt {} }$	${\color{none} \large \texttt {3}}$	${\color{none} \large \texttt {0}}$	15s	$0.00
test_send_native_sequential	${\color{none} \large \texttt {100} \normalsize \texttt {} }$	${\color{none} \large \texttt {3}}$	${\color{none} \large \texttt {0}}$	20s	$0.01

Total run time: 5.37 minutes