fix(ecdatakit): print marker record describing max num. of columns (#424

)

## Description

ECDataKit proceses csv files using `polars` library. It seems that
`pl.DataFrame.read_csv` deduces number of columns from the first record
read and then if it encouters longer record (with more columns)
it ignores additional values.

I could handle it solely on ECDataKit side by reading the whole data
file, prepending it with such marker record and then overwriting it, but
doing it via ecrs seems easier for now.

examples/jssp/problem/probe.rs

@@ -30,7 +30,12 @@ impl Probe<JsspIndividual> for JsspProbe {
     // iterinfo,<generation>,<eval_time>,<sel_time>,<cross_time>,<mut_time>,<repl_time>,<iter_time>
 
     #[inline]
-    fn on_start(&mut self, _metadata: &ecrs::ga::GAMetadata) {}
+    fn on_start(&mut self, _metadata: &ecrs::ga::GAMetadata) {
+        // This is a marker record for ECDataKit. Since it looks like
+        // polars.DataFrame.read_csv deduces number of columns from the first encoutered
+        // record it leads to crashes when longer records are encountered deeper in the file.
+        info!(target: "csv", "event,col_1,col_2,col_3,col_4,col_5,col_6,col_7");
+    }
 
     fn on_initial_population_created(
         &mut self,