add: test results for performance test cases

- generally quite a bit faster now but there seems to be something still
  potentially wrong when translations and/or lots of choices/itemsets
  are involved, e.g. 10K text = ~5 seconds,
  vs 10K select + 20K choices = ~80 seconds.
- switched to using `convert()` directly since this now accepts
  markdown input, and it avoids potentially confounding effects of
  extra things done in `assertPyxformXform` (such as re-parsing XML).
  - this had a minor impact (approx 10%) on test execution time but
    considerable impact (approx 30%) on memory usage.

tests/test_dynamic_default.py

@@ -10,6 +10,7 @@
 
 import psutil
 from pyxform import utils
+from pyxform.xls2xform import convert
 
 from tests.pyxform_test_case import PyxformTestCase
 from tests.xpath_helpers.choices import xpc
@@ -774,13 +775,14 @@ def test_dynamic_default_performance__time(self):
         """
         Should find the dynamic default check costs little extra relative time large forms.
 
-        Results with Python 3.9.10 on VM with 4CPU 8GB RAM, x questions each, average of
-        10 runs (seconds), with and without the check, per question:
-        | num  | with   | without |
-        |  500 | 0.4599 |  0.4535 |
-        | 1000 | 0.9234 |  0.9195 |
-        | 2000 | 2.1118 |  1.9917 |
-        | 5000 | 4.9563 |  4.8714 |
+        Results with Python 3.10.14 on VM with 2vCPU (i7-7700HQ) 1GB RAM, x questions
+        each, average of 10 runs (seconds), with and without the check, per question:
+        | num   | with   | without | peak RSS MB |
+        |   500 | 0.2415 |  0.2512 |          58 |
+        |  1000 | 0.4754 |  0.5199 |          63 |
+        |  2000 | 0.9866 |  1.2936 |          67 |
+        |  5000 | 3.1041 |  2.7132 |          96 |
+        | 10000 | 5.4795 |  5.3229 |         133 |
         """
         survey_header = """
         | survey |            |          |          |               |
@@ -798,7 +800,7 @@ def run(name, case):
                 results = []
                 while runs < 10:
                     start = perf_counter()
-                    self.assertPyxformXform(md=case)
+                    convert(xlsform=case)
                     results.append(perf_counter() - start)
                     runs += 1
                 print(name, round(sum(results) / len(results), 4))

tests/test_translations.py

@@ -1,7 +1,7 @@
 """
 Test translations syntax.
 """
-
+from pyxform.xls2xform import convert
 import unittest
 from dataclasses import dataclass
 from time import perf_counter
@@ -396,13 +396,15 @@ def test_missing_translations_check_performance(self):
         """
         Should find the translations check costs a fraction of a second for large forms.
 
-        Results with Python 3.10.14 on VM with 2vCPU (i7-7700HQ) 4GB RAM, x questions
+        Results with Python 3.10.14 on VM with 2vCPU (i7-7700HQ) 1GB RAM, x questions
         with 2 choices each, average of 10 runs (seconds), with and without the check,
         per question:
-        | num  | with   | without |
-        |  500 | 3.0420 |  3.0427 |
-        | 1000 | 9.7641 |  9.6972 |
-        | 2000 | 30.645 |  28.869 |
+        | num   | with   | without | peak RSS MB |
+        |   500 | 1.0235 |  0.9831 |          74 |
+        |  1000 | 2.3025 |  2.6332 |         101 |
+        |  2000 | 5.6960 |  6.2805 |         157 |
+        |  5000 | 23.439 |  25.327 |         265 |
+        | 10000 | 80.396 |  75.165 |         480 |
         """
         survey_header = """
         | survey |                 |        |                    |                   |
@@ -419,7 +421,7 @@ def test_missing_translations_check_performance(self):
         |         | c{i}        | na   | la-d  | la-e       |
         |         | c{i}        | nb   | lb-d  | lb-e       |
         """
-        for count in (500, 1000, 2000):
+        for count in (10000, ):
             questions = "\n".join(question.format(i=i) for i in range(1, count))
             choice_lists = "\n".join(choice_list.format(i=i) for i in range(1, count))
             md = "".join((survey_header, questions, choices_header, choice_lists))
@@ -429,10 +431,10 @@ def run(name, case):
                 results = []
                 while runs < 10:
                     start = perf_counter()
-                    self.assertPyxformXform(md=case)
+                    convert(xlsform=case)
                     results.append(perf_counter() - start)
                     runs += 1
-                print(name, sum(results) / len(results))
+                print(name, round(sum(results) / len(results), 4))
 
             run(name=f"questions={count}, with check (seconds):", case=md)