In principle there is no physical limitation, you should be able to use CW pumping, but the reality is that to achieve a pump energy density high enough you would surpass the damage threshold of most media. Pulse excitation is more practical.
According to the general theory of lasers, gain threshold is equal to sum of all the losses i. e,
"Gain threshold= Threshold of population inversion*Stimulated emission cross-section = absorption/scattering losses + coupling loss"
Dye solutions (gain medium of the random laser), are known to be as a high gain media. Moreover, Random laser medium includes highly scatterer particles which mean that absorption/scattering losses is also high. So, for such a high inverted population threshold, high pumping power is needed that is supplied using pulsed lasers with intense peak power.
Pulsed lasers with ~10 ns duration and several mJ energy per pulse exhibit several MW peak power, sufficient to achieve sufficient pumping of random lasers while there is no CW laser with such a high power in practice (i.e, MW -CW).
The concept of damage threshold in a short time duration are the same for pulsed and CW laser, while it would be crucial at duration longer than microseconds where serious thermal effect may take place.
In particular case of the chaotic lasing with laser dyes the possible answer on the given quastion is the same as for the case for traditional dye lasing - abberations of gain media due to heating which have been removed by application of dye stream laser version. Abberations in the case of chaotic lasing is not importent, but dye heating with quantum yied decrease and triplet state population provide growing losses and threshold of superfluorescence (ASE) in the case of CW pumping
Yes, it is possible. You can see early papers on electrically pumped RL or a very recent publication (Nanoscale, 2017) using clorophyl based RL with silver NP as scatters pumped at 632nm by a He-Ne cw laser.
To my opinion CW-lasing in the case of random lasing and many other types of laser systems is allowed principally always and every time when the initial threshold conditions do not change in the worse side. Surveillance to stability of threshold conditions comes from heating of laser media due to that laser efficiency is always lower 100%. So competition of cooling against heating determines the life time of lasing time of life. The known method of CW lasing for liquid dye laser had been realized in so called continuous stream dye lasers with CW Argon pumping.
Using a pulsed laser is more practical, particularly in media with fast relaxation time (ZnO is a popular example), as this allows you to bring the laser above threshold before the inverted population naturally relaxes back to the ground state. In theory, you can do this with a CW laser as long as the incoming photon flux is in equilibrium with the emitted photon flux + scattering away from the sample + inversion relaxation + absorptive losses when the gain and absorption bands overlap (rhodamine is one example).
by using CW laser the FWHM reduced slightly and does not appear any spikes at the top of the intensity, but with a pulsed laser the FWHM decreases significantly and also shows a large number of spikes at the top of the intensity and each spike represents a pattern through the energy transfer between the dye molecules, Which we can't reach these spikes by CW laser
High repetition rate pumping may look continuous. However, the scattering process itself occurs at time intervals that make the fundamental action pulsed.
I'd like to add, there are recent reports of random lasing with CW pumping in fiber random lasers. Photon confinement in the fiber core creates a partially closed cavity, which reduces net losses in the system. This then equates to a lower threshold, and the laser can be brought above threshold with a weaker pump source.