Parameters provided in examples (and default parameters) does not perform BOLD simulation correctly

[yusenz]

For some reason I do not understand, the append_outputs parameter in model.run() is defaulted to None and thus False in model.integrateChunkwise(). Meanwhile, the bold simulation in model.simulateBold() pulls from self.default_output.
As a result, the BOLD simulation, when running with default parameters, will ONLY use the simulated activity of last chunk to simulate the entire duration. This is especially noticable when the model is run with time-dependent external stimuli.

The examples with BOLD like example 0 aln model suffer from this issue, and it has a strong misleading effect.

[yusenz]

Edit: setting append_outputs=True does not fix the issue. The bold activity still behaves like the stimuli is not time dependent at all. The previous comment on the cause of the issue might be wrong. For my experiments setting chunkwise=False has acceptable memory cost and fixed the problem. However this needs further inspection from the team.