error: run rl examples

🐛 Bug Description

To Reproduce

Steps to reproduce the behavior:

1. test rl according to examples/rl/README.md

2.      File "train_onpolicy.py", line 79, in __getitem__
            self._ticks_index = [t - date for t in backtest_data.get_time_index()]
          File "train_onpolicy.py", line 79, in <listcomp>
            self._ticks_index = [t - date for t in backtest_data.get_time_index()]
        TypeError: unsupported operand type(s) for -: 'tuple' and 'Timestamp'
   

Environment

          Linux
          x86_64
          Linux-5.10.102.1-microsoft-standard-WSL2-x86_64-with-glibc2.29
          #1 SMP Wed Mar 2 00:30:59 UTC 2022
          
          Python version: 3.8.5 (default, May 27 2021, 13:30:53)  [GCC 9.3.0]
          
          Qlib version: 0.8.6.99
          numpy==1.20.3
          pandas==1.3.5
          scipy==1.9.3
          requests==2.28.1
          sacred==0.8.2
          python-socketio==5.7.2
          redis==4.3.5
          python-redis-lock==4.0.0
          schedule==1.1.0
          cvxpy==1.2.2
          hyperopt==0.1.2
          fire==0.4.0
          statsmodels==0.13.5
          xlrd==2.0.1
          plotly==5.11.0
          matplotlib==3.6.2
          tables==3.7.0
          pyyaml==5.3.1
          mlflow==1.30.0
          tqdm==4.64.1
          loguru==0.6.0
          lightgbm==3.3.3
          tornado==6.2
          joblib==1.0.0
          fire==0.4.0
          ruamel.yaml==0.17.21

Start with examples/rl/README.md and reproduce the RL example on the reported environment. Inspect train_onpolicy.py at line 79 and trace how backtest_data.get_time_index() is produced; done means the example runs past this TypeError with its time index handling verified by the relevant RL example or tests.