[Code scan] Replace ineffective assertTrue checks with equality assertions

[njzjz]

This issue is from a Codex global scan of the repository.

Two tests use assertTrue(actual, expected) where the second argument is only the failure message, so the intended comparison is never performed.

Evidence:

dpgen/tests/generator/test_make_train.py

Lines 59 to 63 in 7af5246

	def _check_numb_models(testCase, iter_idx, numb_models):
	models = glob.glob(
	os.path.join("iter.%06d" % iter_idx, "00.train", "[0-9][0-9][0-9]") # noqa: UP031
	)
	testCase.assertTrue(len(models), numb_models)

dpgen/tests/auto_test/test_lammps.py

Lines 75 to 82 in 7af5246

	def test_make_potential_files(self):
	cwd = os.getcwd()
	abs_equi_path = os.path.abspath(self.equi_path)
	self.Lammps.make_potential_files(abs_equi_path)
	self.assertTrue(os.path.islink(os.path.join(self.equi_path, "frozen_model.pb")))
	self.assertTrue(os.path.isfile(os.path.join(self.equi_path, "inter.json")))
	ret = loadfn(os.path.join(self.equi_path, "inter.json"))
	self.assertTrue(self.inter_param, ret)

_check_numb_models() currently passes whenever len(models) is nonzero, even if it does not equal numb_models. test_make_potential_files() passes whenever self.inter_param is truthy; the loaded ret value is used only as a message and is not compared.

Expected behavior: use equality assertions such as assertEqual(len(models), numb_models) and assertEqual(ret, self.inter_param) so these tests catch regressions.