[OpenCL] Better graphs for profiling results ...

[kif]

def log_profile(self):
        """If we are in profiling mode, prints out all timing for every single OpenCL call
        
        :param stats: if True, prints the statistics on each kernel instead of all execution timings
        :return: list of lines to print
        """
        total_time = 0.0
        stats = {}

        if self.profile:
            for e in self.events:
                if isinstance(e, ProfileDescription):
                    name = e[0]
                    t0 = e[1]
                    t1 = e[2]
                elif isinstance(e, EventDescription) or "__len__" in dir(e) and len(e) == 2:
                    name = e[0]
                    pr = e[1].profile
                    t0 = pr.start
                    t1 = pr.end
                else:
                    name = "?"
                    t0 = e.profile.start
                    t1 = e.profile.end

                et = 1e-6 * (t1 - t0)
                total_time += et
                if name in stats:
                        stats[name].append(et)
                else:
                        stats[name] = [et]
        return stats
log_profile(gpu)
from silx.opencl.processing import EventDescription, ProfileDescription
log_profile(gpu)
stats = log_profile(gpu)
import matplotlib.cbook as cbook
list(stats.keys())
import numpy
ary = numpy.zeros((len(stats),max(len(i) for i in stats.values())))
ary.shape
for i,k in enumerate(stats):
    ary[i] = stats[k]
import matplotlib.cbook as cbook
import matplotlib.pyplot as plt
fig, axs = plt.subplots()
axs.bxp(bstats[-1::-1], vert=False)
axs.set_xlabel("Runtime (ms)")
fig.suptitle("Runtime of different kernels Nvidia A5000")
fig.suptitle("Runtime of different kernels on Nvidia A5000")
axs.set_title("Average runtime timeit: 2.7ms, OpenCL: 2.2ms")

[kif]

Issue: what if the number of calls to the kernel differ ... use NaN ?