Having separate APIs means that if we want the segmentation data corresponding to the point cloud, it has to be called just after getLidarData
Even this is not foolproof, and might get mismatched data. Related issue #2409
Combining the 2 APIs seemed to be to a cleaner solution. Also add generic findSensorByName function in VehicleApiBase

A simple test-script -

import setup_path
import airsim
import time

client = airsim.CarClient()
client.confirmConnection()

count = 0

for i in range(1000):
    lidarData = client.getLidarData()
    # lidarSegData = client.simGetLidarSegmentation()
    lidarSegData = lidarData.segmentation
    
    if not lidarData.point_cloud:
        print("Null entry, skipping")
        continue

    print(f"Lidar: {len(lidarData.point_cloud)}")
    # time.sleep(0.1)
    print(f"LidarSeg: {len(lidarSegData)}")

    if(len(lidarSegData) != len(lidarData.point_cloud)//3):
        print("Not equal")
        count+=1
    time.sleep(0.1)

print(count)

For envs like Neighbourhood, I'm getting 30-40 out of 1000 mismatched entries (which isn't a lot actually)

Note: I haven't added the workaround for empty segmentation vecotr similar to what is there for point_cloud in RpcLibAdaptorsBase since rpclib/rpclib#152 has been fixed. I've opened #2812 for removing the workaround, hopefully they rebase cleanly
With the removal of the workaround, checking whether the point_cloud is empty using if not lidarData.point_cloud: works

Edit: Well, after thinking about it a bit more, it somewhat makes sense that the segmentation API is a different one since this data won't actually be present in real life, therefore sim prefix. This problem can also be detected pretty easily, so if we want to keep the APIs separate, the first option could be to close this PR (or just use the last commit)
Another could be to create a new Lidar sensor which has the segmentation info as well. This will also lead to the removal of collecting segmentation data in the normal Lidar sensor, which probably will improve performance (but will need to check how much it's affected)
Let me know whatever seems best