N-M calculation discrepancies in cross section

[DanielGMorenaFhecor]

Contrasting the results from this library with proven software, the N-M diagram results are inconsistent. When the center of gravity is at the origin, the results are more accurate. It would also be beneficial to obtain the full N-M diagram, not just the lower range of M.

See Point 10 that @MestreCarlos prepared in this notebook for further information and illustrations.

[talledodiego]

For the first part, see also #140 and #141 that I think explains the apparent discrepancies.

For the second part: I remember we have spent some time thinking about it with @mortenengen (there is still a notebook somewhere with my comment saying # Morten: two calls? Or a single call? 🤣) Since in the API for calculate_nm_interaction_domain we included the angle theta of the neutral axis, it makes sense that only half is drawn sinche with theta equal to pi the other part is returned.

Something like this:

# Morten: two calls? Or a single call?
res_1 = section.section_analyzer.calculate_nm_interaction_domain(theta=0)
res_2 = section.section_analyzer.calculate_nm_interaction_domain(theta=0+np.pi)

fig_mn, ax_mn = plt.subplots()
ax_mn.plot(res_1.n / 1000, res_1.m_y *1e-6)
ax_mn.plot(res_2.n / 1000, res_2.m_y *1e-6)
ax_mn.set_xlabel('N [kN]')
ax_mn.set_ylabel('M [kNm]')
ax_mn.axvline(0.0, linestyle='--', color ='k')
ax_mn.axhline(0.0, linestyle='--', color ='k')
ax_mn.set_xlim([-6500, 1500])
ax_mn.xaxis.set_inverted(True)

i_begin = 0
colors = colormaps['tab20']
j = 2
for i in (3,10,40,10,40,40):
    i_end = i_begin + i + 1

    ax_mn.plot(res_1.n[i_begin:i_end] / 1000,res_1.m_y[i_begin:i_end] *1e-6,'-o',markersize=2,color=colors(j))

    ax_mn.plot(res_2.n[i_begin:i_end] / 1000,res_2.m_y[i_begin:i_end] *1e-6,'-o',markersize=2,color=colors(j+1))
    ax_mn.plot([0, res_2.n[i_begin:i_end][-1]*1e-3], [0, res_2.m_y[i_begin:i_end][-1]*1e-6], '-k')
    ax_mn.plot([0, res_1.n[i_begin:i_end][-1]*1e-3], [0, res_1.m_y[i_begin:i_end][-1]*1e-6], '-k')
    i_begin = i_end - 1
    j+=2

[mortenengen]

I see the point of returning the full domain in one call, but can we do the following:

• Return only the domain related to the negative moment, as today.
• Explicitly state in the docstring that this is the intended behaviour?
• Include examples in the docs where the full domain is calculated using the method shown by @talledodiego above?