Use pgrep to find the redpanda pid

It seems like redpanda_pid can pick up false pids in the current grep
filtering approach. Use `pgrep` to not incorrectly pickup a command with
a command line arg of `redpanda`, which seems likely in our environments.

When debugging redpanda-data#11275, it seems like there is an apport process who's
pid is being picked up. Here's a process dump that includes the line.
There are no running redpanda processes, but checking for redpanda
processes times out, leading me to believe that this process is being
picked up.

In that case `stop_node` can fail if the node is already failed in CDT,
as this process is picked up.

```
[DEBUG - 2023-06-12 02:54:07,216 - redpanda - _log_node_process_state - lineno:2215]: root      136962 99.6  6.0 1587948 1452804 ?     R    02:53   0:49 /usr/bin/python3 /usr/shar
e/apport/apport -p136954 -s5 -c18446744073709551615 -d1 -P136954 -u0 -g0 -- !opt!redpanda_installs!head!libexec!redpanda
```

Fixes: redpanda-data#11275

Signed-off-by: Tyler Rockwood <[email protected]>
(cherry picked from commit 4a2985f)

tests/rptest/services/redpanda.py

@@ -2083,7 +2083,8 @@ def redpanda_pid(self, node):
         # we need to look for redpanda pid. pids() method returns pids of both
         # nodejs server and redpanda
         try:
-            cmd = "ps ax | grep -i 'redpanda' | grep -v grep | grep -v 'version'| grep -v \"\[redpanda\]\" | awk '{print $1}'"
+            cmd = "pgrep --exact redpanda"
+
             for p in node.account.ssh_capture(cmd,
                                               allow_fail=True,
                                               callback=int):