`Lightcuve.plot(witherrors=True)` does not plot error bars

[dhuppenkothen]

Description of the Bug

I got a bug report from a colleague that using the plot function in Lightcurve with the witherrors=True option does not, in fact produce error bars.

I reproduced the issue using the minimum example below.

Steps/Code to Replicate the Bug

time = np.arange(100)
counts = np.random.poisson(400, size=len(time))
lc = Lightcurve(time, counts, err_dist="poisson")

lc.plot(witherrors=True)

Expected Results

I expected something akin to this:

Actual Results

The code produces the following plot, which is not what we'd expect.

I suspect something might have gone wrong when that functionality moved to the StingrayTimeseries superclass and somehow the if-statement that would produce the error bars no longer is executed correctly.

[pupperemeritus]

I think its to do with the initialization of counts_err. I added a check in base.StingrayObject.plot() and it did raise the error when I tested with it.

        if witherrors and attr + "_err" in self.array_attrs():
            ax.errorbar(
                self.time,
                getattr(self, attr),
                yerr=getattr(self, attr + "_err"),
                fmt="o",
                zorder=10,
            )
        elif witherrors and attr + "_err" not in self.array_attrs():
            raise ValueError(f"Did not initialize {attr} + _err in array_attrs")

Traceback (most recent call last):
  File "/home/pupperemeritus/testing_stingray/fixing_plot_err_bug.py", line 9, in <module>
    lc.plot(witherrors=True)
  File "/home/pupperemeritus/stingray/stingray/lightcurve.py", line 1638, in plot
    return super().plot(
           ^^^^^^^^^^^^^
  File "/home/pupperemeritus/stingray/stingray/base.py", line 2457, in plot
    raise ValueError(f"Did not initialize {attr} + _err in array_attrs")
ValueError: Did not initialize counts + _err in array_attrs

When I check if lc.counts_err is accessible, it is. But it fails to materialize in the plot function. There's missing plumbing to get the counts_err through to StingrayTimeseries.plot.