make_package with split=True fails for trivial examples

[spj101]

Reported to me by: Matteo Marcoli

The code generation for simple integrals with make_package and split=True fails.

Example:

from pySecDec import make_package

if __name__ == "__main__":

   make_package(
       name = 'A30b',
       integration_variables = ['z'],
       prefactor = '1',
       regulators = ['eps'],
       requested_orders = [0],
       polynomials_to_decompose = ['(1-z)^(-1-eps)'],
       split=True,
       
       # Bug in symmetry finder in pySecDec v1.6.4
       use_iterative_sort=True, # works if set to False
       use_light_Pak=False,
       use_dreadnaut=False,
       use_Pak=False,
   )

Gives:

running "sum_package" for A30b
running "make_package" for "A30b_integral"
number of primary sectors before symmetry finding: 1
Traceback (most recent call last):
  File "/Users/sj/Documents/repo/secdec-private/nodist_examples/marcoli/makepkg.py", line 5, in <module>
    make_package(
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/make_package.py", line 375, in make_package
    sum_package(
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/code_writer/sum_package.py", line 480, in sum_package
    template_replacements = [_generate_one_term(
                            ^^^^^^^^^^^^^^^^^^^^
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/code_writer/sum_package.py", line 480, in <listcomp>
    template_replacements = [_generate_one_term(
                             ^^^^^^^^^^^^^^^^^^^
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/code_writer/sum_package.py", line 171, in _generate_one_term
    mp = make_package(**package_generator._asdict())
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/code_writer/make_package.py", line 1988, in make_package
    for primary_sector_index, primary_sector in enumerate(primary_sectors_to_consider):
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/code_writer/make_package.py", line 1856, in primary_decomposition_with_splitting
    primary_sectors = _reduce_sectors_by_symmetries\
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/code_writer/make_package.py", line 1432, in _reduce_sectors_by_symmetries
    sectors = decomposition.squash_symmetry_redundant_sectors_sort(sectors, iterative_sort, indices)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/decomposition/common.py", line 658, in squash_symmetry_redundant_sectors_sort
    replaced_sectors = _remove_variables(sectors, indices)
                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/decomposition/common.py", line 575, in _remove_variables
    output.append( Sector(new_cast, new_other, new_Jacobian) )
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/decomposition/common.py", line 53, in __init__
    initial_monomial_factor = Polynomial([ [0]*self.number_of_variables ], [1], poly.polysymbols)
                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/sj/Documents/repo/secdec-private/pySecDec/algebra.py", line 381, in __init__
    raise TypeError('All entries in `expolist` must be integer.')
TypeError: All entries in `expolist` must be integer.

The problem seems to be that split is implemented as a kind of primary decomposition and the symmetry finder ends up trying to (incorrectly) create a Sector that has the 1 variable z eliminated.

Disabling the symmetry finder gives reasonable looking results.