Vs_Profile(add_halfspace=True) leaves material number 0 on the old last layer, so the new profile fails the class's own format check #106

Found while reviewing #74. Pre-existing on main (b764057); #74 keeps the same behavior.

Problem

check_Vs_profile_format() allows material number 0 on the last layer, whatever its thickness, and requires a positive material number on every other layer. So a 5-column profile whose last layer is finite (thickness > 0) and has material number 0 is accepted. With add_halfspace=True, Vs_Profile() appends a half-space row with material number 0 below that layer, but leaves the old last layer at 0. That layer is no longer the last one, so prof.vs_profile breaks the rule above: check_Vs_profile_format() rejects it, and Vs_Profile(prof.vs_profile) raises ValueError.

truncate() already handles the same case: when the layer above the new half-space has material number 0, it gives that layer a new number, max + 1:

Also: the error message has a typo. It says "except for the last error" instead of "except for the last layer" (line 540 above). The repro shows the message.

import numpy as np
import PySeismoSoil.helper_generic as hlp
from PySeismoSoil.class_Vs_profile import Vs_Profile

data = np.array([
    [10.0, 200.0, 0.05, 1800.0, 1.0],
    [20.0, 300.0, 0.05, 1800.0, 0.0],  # last layer: finite, material number 0
])
hlp.check_Vs_profile_format(data)  # accepted

prof = Vs_Profile(data, add_halfspace=True)
print(prof.vs_profile[:, [0, 4]])
# [[10.  1.]
#  [20.  0.]
#  [ 0.  0.]]
# (expected [[10.  1.]
#            [20.  2.]
#            [ 0.  0.]])

try:
    hlp.check_Vs_profile_format(prof.vs_profile)
except ValueError as e:
    print(type(e).__name__, e)
# ValueError The "material number" column should be all positive, except for the last error.
# (expected: no error)

try:
    Vs_Profile(prof.vs_profile)
except ValueError as e:
    print(type(e).__name__, e)
# ValueError The "material number" column should be all positive, except for the last error.
# (expected: no error)

# truncate() handles the same case by giving the layer a new material number
print(Vs_Profile(data).truncate(depth=40).vs_profile[:, [0, 4]])
# [[10.  1.]
#  [30.  2.]
#  [ 0.  0.]]

(truncate() also changes the original profile in place, see #75. That does not affect the output above.)

The existing tests test_add_halfspace__case_1__already_a_half_space and test_add_halfspace__case_1__no_half_space on main (in #74: test_add_halfspace[already_a_half_space] and test_add_halfspace[no_half_space]) do not cover this case. The first file already has a half-space, and the second is a 2-column profile, which gets material numbers 1..n.

Fix

1. In Vs_Profile.__init__, before appending the half-space, give the old last layer a new material number when it has 0, as truncate() does:

               half_space[4] = 0  # set material number to 0
   +            if full_data[-1, 4] == 0:  # the old last layer needs a material
   +                full_data[-1, 4] = np.max(full_data[:, 4]) + 1
   +
               full_data = np.vstack((full_data, half_space))

2. In check_Vs_profile_format(), change "except for the last error." to "except for the last layer.". The material_number_not_positive case of test_Vs_profile_format (in Combine near-duplicate test functions into parametrized tests #74; test_Vs_profile_format__case_10 on main) matches only 'should be all positive', so it still passes.

3. Add a case to test_add_halfspace for a 5-column profile without a half-space whose last layer has material number 0. Check that the material numbers become [1, 2, 0] and that Vs_Profile(prof.vs_profile) works.

With only steps 1 and 2 on main, the repro gives the expected output, and all 240 tests pass.