Improve error/warning messages concerning energy barriers in Arkane

[alongd]

Added a warning if a reaction barrier is too high in Arkane.
Also enhanced a related error message in tunnelling.
Minor: corrected reciprocal T axis units in Arkane plots (1000/K is strange, should be either 1/K or K^-1)

[alongd]

Addresses #1574

[codecov]

Codecov Report

Merging #1575 into master will not change coverage.
The diff coverage is n/a.

@@          Coverage Diff           @@
##           master   #1575   +/-   ##
======================================
  Coverage    41.5%   41.5%           
======================================
  Files         176     176           
  Lines       29306   29306           
  Branches     6033    6033           
======================================
  Hits        12163   12163           
  Misses      16292   16292           
  Partials      851     851

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 85c81ea...3921d0c. Read the comment docs.

[cjmcgill]

The 1000/K axis is a pretty common convention to get the axis numbers into a more readily readable range, numbers in the ones or tenths digit instead of being in the thousandths digit and a bunch of leading zeros. The 1000/K axis is strangely artificial though and even though it makes the results more readable, it does not make them any more intelligible.

Might I suggest instead adding a second axis, either in K or C? Something like what this link describes doing in Origin. Then you can keep either 1000/K or 1/K on the primary axis to keep the linearity of the Arrhenius plot, but the secondary axis makes it more easily understood which is probably your goal.

http://blog.originlab.com/graphing/display-celsius-as-top-axis-in-arrhenius-plot

[alongd]

Thanks for the suggestion, I agree it looks better to have a dual T axis.
I'll clarify that my original comment was just for the axis units (should be 1/K and not 1000/K as we currently have), I did not mean to change the axis scaling (we'll keep the primary axis at 1000/T, not 1/T).
Again, thanks for the helpful comment, I'll try to implement it!

[cjmcgill]

Oh I see! Yeah, having the units be 1000/K makes it ambiguous if the transform has already been done or done twice. Weird minor error. Carry on!

[alongd]

@goldmanm, could you take a look?

[alongd]

@goldmanm, could you take a look at this PR?

[alongd]

@goldmanm, could you take a look at this PR?

[goldmanm]

@alongd, I don't know how I missed the messages before. Looking at it now.