There seem to be many steps required to set the stage for endosymbiosis. There’s the initial interaction with bacteria that includes the steps for envelopment. There are probably a number of steps in the initial association (pre-permanent endosymbiosis) which affect whether the interactions are mutualistic or not. And there is prevention of lyosome fusion. So, there are probably a number of steps leading up to that transition.
As for plastid evolution, we see cases where secondary and tertiary acquisition seems to have occurred in several instances (endosymbiosis of eukaryotic cells containing chloroplasts) . There are also cases where photosynthetic bacteria are temporarily sequestered by hosts (see also, “non-obligate endosymbiosis”). And there are examples of kleptoplasty in which the plastids are removed and retained by the host.
So yes, there are distinct transitions that must occur in obligate endosymbiosis but there must also be stages leading up and supporting the transition.
That mitochondria are derived from an endosymbiotic event is largely a closed mystery (circa early 1990s’) for which evidence continues to accumulate.
There are examples of prokaryotes living inside other prokaryotes. Though we see a single lineage today, we’re not sure if the eukaryotes branched before or after acquisition of the mitochrondrion. Eukaryotes display a number of features that distinguish the group from the archaea and bacteria. However, it seems that the plastids became endosymbionts after eukaryotes emerged.
Additionally, we have evidence of a novel acquisition of a plastid via endosymbiosis with a cyanobacterium by an amoeba about 60 mya (Paulinella chromatophora).