Improve diagnostics & reporting of lineup2() plan

[prjemian]

• close lineup2() plan needs better diagnostics, especially when it does not succeed #1044
  @jilavsky : Thanks for the prompting on this!

[jilavsky]

That looks great, thank you

[prjemian]

success=True

Motor: 'motor'  Detector: 'noisy_det'
========== ====================
statistic  value               
========== ====================
n          19                  
centroid   -0.00604313560021479
x_at_max_y 0.07498907624123874 
fwhm       2.427762264595814   
variance   0.8626928293778106  
sigma      0.9288125910956475  
min_x      -3.0                
mean_x     0.9535573837387351  
max_x      4.90711476747747    
min_y      -0.08111386966031231
mean_y     0.29494471116616    
max_y      0.980899512651778   
success    True                
========== ====================

success=False

Motor: 'motor'  Detector: 'noisy_det'
========== ================================
statistic  value                          
========== ================================
n          19                             
centroid   -0.5391606368907506            
x_at_max_y 0.0                            
fwhm       3.0223874138499687             
variance   0.0                            
sigma      0.0                            
min_x      -3.0                           
mean_x     1.5                            
max_x      6.0                            
min_y      -0.09436187725503624           
mean_y     0.19753874119348702            
max_y      0.7724043891492777             
success    False                          
reasons    ['Computed peak sigma is zero.']
========== ================================

[prjemian]

To get the peak analysis into the dictionary on a routine basis, need to insert a call for that analysis before the call to self.report():

apstools/apstools/callbacks/scan_signal_statistics.py

Lines 199 to 205 in bb907b9

	self.analysis = array_statistics(
	self._data[self._x_name],
	self._data[self._detectors[0]],
	)
	if self.reporting:
	self.report()

How to keep SignalStatsCallback generic? Move the peak analysis code into the callback class? Then refactor the success key into feature_found?

[prjemian]

Each of these inner functions rely on content in SignalStatsCallback.analysis (a MMap structure). They could be moved into utils.statistics and called from a re-factored SignalStatsCallback. The unit tests would better target each of these these function's results.

apstools/apstools/plans/alignment.py

Line 503 in bb907b9

def find_peak_position():

apstools/apstools/plans/alignment.py

Line 536 in bb907b9

def strong_peak(stats) -> bool:

apstools/apstools/plans/alignment.py

Line 546 in bb907b9

def too_wide(stats):

With that refactor, this decision would examine the appropriate SignalStatsCallback.analysis key:

apstools/apstools/plans/alignment.py

Line 574 in bb907b9

if target is None:

[prjemian]

@MDecarabas Thanks!