It is not very useful to measure the temperature of the motor steel case, because that's the stator, and it is not touching the rotor where the windings are. So the rotor can be smoking, while the stator is still cool.
However, your motor comes with a built-in thermometer, because the temperature coefficient resistance of its copper windings is 0.39%/°C.
Therefore, connect the motor to a bench power supply and stall it. Note the current going through it, then wait for it to heat.
As copper gets hotter, its resistance increases. So you should see current decrease. When it has decreased by 39%, you know the winding temperature has increased by 100°C. So if it was 20°C at the beginning, it is now 120°C, and it is time to disconnect the motor to avoid burning it. The steel case should still be cool, but it'll be cooking inside.
Note you don't need to do this experiment, you can just calculate the hot resistance value.
You can exploit this to measure the temperature of your motor and how much power it can dissipate while stalled.
Let it cool down. Then, you can set the bench power supply to 0V, stall the motor again, and slowly increase the voltage, in small steps.
Once the voltage has settled,
if the motor receives more heating power than it can dissipate in the air, its temperature will increase, so you will see the current decrease slowly while voltage remains constant.
On the other hand, if the motor receives less heating than it can dissipate in the air, it will cool down, so current will increase slowly.
You are looking for a voltage setting that:
keeps the current stable, which means the temperature is constant, which allows you to determine the power dissipated in the air, which will be equal to the input electrical power
keeps the motor's resistance about equal to the value measured previously, corresponding to a temperature rise of 100°C (you can use a lower temperature rise if it begins to smell funny, but 100°C should be okay). Just divide V by I displayed by the bench power supply to get the resistance, and you know the winding temperature.
The settings on the power supply will be the maximum power your motor can dissipate while stalled. You can translate that into PWM settings, and check with the multimeter that V*I is indeed what you wanted.
If you don't get enough torque, another solution is to use a fan to blow some air into the motor when it is stalled. Not on the motor, but into it, through the vents, because you want to cool the windings. When it is turning, its own fan works, but not when it is stalled. That is not likely to perform any miracle, but who knows.