`TropCyclone.from_tracks` creates invalid intensity matrix for duplicate coordinates

[sarah-hlsn]

Describe the bug
When centroids are generated from exposure objects containing duplicates, the resulting hazard intensity matrix contains negative indices:
When running tc_pnt.intensity.indices for the example below, it returns array([ 0, 1, 2, 4, -1, 5], dtype=int32)

To Reproduce
see example

Code example:

import numpy as np
from climada.entity import Exposures
from climada.hazard import Centroids, TCTracks, TropCyclone

latitude = np.array(
    [21.899326, 21.960728, 22.220574, 22.298390, 21.787977, 21.787977, 21.981732]
)
longitude = np.array(
    [88.307422, 88.565362, 88.378337, 87.806356, 88.348835, 88.348835, 89.246521]
)

values = np.array([1.0e5, 1.2e5, 1.1e5,1.1e5, 2.0e5, 2.5e5, 0.5e5])

exp_pnt = Exposures(
    lat=latitude,  
    lon=longitude,  
    value=values,
    value_unit="USD",
    crs="epsg:4326")

centr_pnt = Centroids.from_exposures(exp_pnt)

tr_pnt = TCTracks.from_ibtracs_netcdf(storm_id="2007314N10093")
tc_pnt = TropCyclone.from_tracks(tr_pnt, centroids=centr_pnt)
tc_pnt.check()
tc_pnt.intensity.indices

Climada Version: 6.0.2.dev

System Information:

• Operating system and version: macOS 14.0
• Python version: 3.11.11

[peanutfun]

This is not exclusive to the centroids created from exposure, but also occurs when centroids are created from coordinates:

import numpy as np
from climada.hazard import Centroids, TCTracks, TropCyclone

# NOTE: Duplicate coordinates at indices 4, 5
latitude = np.array(
    [21.899326, 21.960728, 22.220574, 22.298390, 21.787977, 21.787977, 21.981732]
)
longitude = np.array(
    [88.307422, 88.565362, 88.378337, 87.806356, 88.348835, 88.348835, 89.246521]
)

centr_pnt = Centroids.from_lat_lon(latitude, longitude)

tr_pnt = TCTracks.from_ibtracs_netcdf(storm_id="2007314N10093")
tc_pnt = TropCyclone.from_tracks(tr_pnt, centroids=centr_pnt)
tc_pnt.check()
tc_pnt.intensity.indices