The power limits you are quoting are thermal limits. I.e the amount of heat generated by that power level, above which voice coil will be damaged through over heating
Excursion limit is frequency dependent. Above a certain frequency thermal limit takes over but below that less power than the thermal limit will actually cause the driver to exceed xmax. Dependent on driver the xlim may just be beyond or way beyond xmax. The spider usually limits travel but depending on backplate design the coil can smack the backplate.
For a 3: way speaker a rule of thumb is 10/30/60 % power spread across T/M/W drivers
It depends on your average volume level too. Allowing for 10-20dB headroom. Means 10 to 100 times reserve power. So an average 1w/1m at 86dB can be power limited at 106dB with a 200w amp
More is better imho. Just use your ears and listen for distortion
Agreed - Before I started a business providing live sound, while in college, I did extensive real life experiments into what would destroy conventional cone shaped speaker drivers. I drove several speaker drivers with bandpassed filtered music and sine waves (while wearing ear protection) to destruction and found the following:
Below the lower midrange, damage from over excursion usually caused the former to hit the backplate (especially if the driver was not loaded properly) jamming the driver in place as the former was damaged/crushed preventing it from moving in the magnetic gap.
The midrange frequencies were actually tolerated the best as the excursions were more limited than in the lower frequency trials, preventing mechanical damage, and the mid frequencies apparently provided more ventilation to the former/voice coil assembly than the high frequencies in the high frequency trials. Beyond a certain level however the glue holding the voice coil on the former would melt, char, and/or vaporize delaminating the voice coil and jamming the former/voice coil assembly in the magnetic gap.
High frequencies (although usually at a lower level in music) when pushed with the same amount of watts as in the mid frequency trials actually damaged the drivers sooner than mid frequencies (although with similar damage) apparently because driver motion was rather limited and did not provide any cooling to the voice coil and former. (If I had used tweeters instead of cone drivers there likely would also have been other types of damage such as the coil actually melting as tweeter wire is significantly thinner - however my experiments were limited to cone drivers.)
Although I conducted these experiments many years ago I clearly recall the equipment used: Soundcraftmen MA-5002 amplifier, (with crowbar protection circuit), Crown VFX2A crossover, Moog Minimoog synthesizer, and several scavenged cone drivers varying between 4" to 8".
I also recall that all these blisteringly loud trials made me rather unpopular with my neighbors on that summer Sunday afternoon.