Invalid points in LaserScan aren't handled correctly with "apply_correction"

[kawk]

With R2300 4-layer and active "apply_correction", malloc() errors occur if the original LaserScan data contains NaN or Inf values. These can occur since #18 which was fixed only a few weeks before.

To Reproduce
Prepare R2300 at 10.0.10.76 and start driver with /tmp/params.yaml as below

ros_main --ros-args -r __node:=ros_main --params-file /tmp/params.yaml

ros_main:
  ros__parameters:
    transport: "udp"
    scanner_ip: "10.0.10.76"
    max_num_points_scan: 0
    frame_id: "scanner_link"
    start_angle: -450000
    packet_type: "C1"
    scan_topic: "/cloud"
    num_layers: 4
    apply_correction: true
    watchdogtimeout: 60000
    watchdog: false

Symptoms

As soon as the driver receives scan data from sensor containing invalid values, a malloc() error like this will occur:

...
[INFO] [1736171940.198833507] [ros_main]: Device state changed to Running
malloc(): corrupted top size

Experienced with ROS Humble.

[kawk]

The code doing the "correction" of points from LaserScan into PointCloud incorrectly does not skip points where range is NaN or Inf (see comments in #23 for details).

[kawk]

Fixed in #23