[tvla] Enable trace filtering for TVLA-general

Tvla has a mechanism for filtering out outlier traces from the
project. At the moment, this filtering is applied only to
aes specific TVLA. This commit enables filtering for general TVLA
as well.
As a result of applied filtering, the expected result of the CI
test test_general_aes_nonleaking_project() has changed. This
commit also updates the expected result of this test, in order
to leave the repository in a working state.

Signed-off-by: Vladimir Rozic <[email protected]>

cw/tvla.py

@@ -483,28 +483,23 @@ def run_tvla(ctx: typer.Context):
                                                          trace_start] + 0.5) * trace_resolution
                     traces = traces.astype('uint16')
 
-                if general_test is False:
-                    # Filter out noisy traces.
-                    log.info("Filtering Traces")
+                # Filter out noisy traces.
+                log.info("Filtering Traces")
 
-                    # Get the mean and standard deviation.
-                    mean = traces.mean(axis=0)
-                    std = traces.std(axis=0)
+                # Get the mean and standard deviation.
+                mean = traces.mean(axis=0)
+                std = traces.std(axis=0)
 
-                    # Define upper and lower limits.
-                    max_trace = mean + num_sigmas * std
-                    min_trace = mean - num_sigmas * std
+                # Define upper and lower limits.
+                max_trace = mean + num_sigmas * std
+                min_trace = mean - num_sigmas * std
 
-                    # Filtering of converted traces (len = num_samples). traces_to_use itself can be
-                    # used to index the entire project file (len >= num_samples).
-                    traces_to_use = np.zeros(len(project.waves), dtype=bool)
-                    traces_to_use[trace_start:trace_end + 1] = np.all((traces >= min_trace) &
+                # Filtering of converted traces (len = num_samples). traces_to_use itself can be
+                # used to index the entire project file (len >= num_samples).
+                traces_to_use = np.zeros(len(project.waves), dtype=bool)
+                traces_to_use[trace_start:trace_end + 1] = np.all((traces >= min_trace) &
                                                                       (traces <= max_trace), axis=1)
-                    traces = traces[traces_to_use[trace_start:trace_end + 1]]
-                else:
-                    # For now, don't perform any filtering when doing general fixed-vs-random TVLA.
-                    traces_to_use = np.zeros(len(project.waves), dtype=bool)
-                    traces_to_use[trace_start:trace_end + 1] = True
+                traces = traces[traces_to_use[trace_start:trace_end + 1]]
 
                 if i_step == 0:
                     # Keep a single trace to create the figures.