Yes they are sufficient and such bare minimum plasmids are available now adays and are popular with DNA synthesis companies. These stretches you mention are due to how the plasmids were derived in earlier days (from natural plasmids) since then they were just looking for something that would work for cloning purposes and is reasonably easy to use. For example pSC101 (one of the first cloning vectors) was derived from the shearing of a mutant form of naturally occurring plasmid called R6 and looking for a mini-fragment capable of replication and can be selection with tertacycline (ref below). This mini-fragment represents 10% of the size of the large parental plasmid R6-5 hence you can see that a large portion of plasmid is not actually needed for the cloning purpose and would become the extra unnecessary sequences. You see that now, especially since the replication and marker regions are much better delimited now as compared to back then. Other reasons are the limitation is cloning ability when creating a vector in the earlier days (only a limited number of restriction enzymes are known) and hence would force you to leave extra sequences even though you do not want it. Many plasmids used now are dervied from these older vectors hence these sequences still persist now. If you are interested in more example your can read about the derivation of the pUC series of plasmids and pBR322.

Article Recircularization and Autonomous Replication of a Sheared R-...

I would say it depends in part if you want the plasmid to be able to move from one cell to another, such as the F plasmid encoding an F pilus to transfer the plasmid to an F- cell. A selection marker is not needed for a wild plasmid, it is only needed for genetic engineering ease.

The only thing you actually need for a plasmid is the origin of replication. You could have a plasmid that is much smaller than 1kb (for example piVX is around 700bp with a selectable marker). You can add a selectable marker if you want to select and maintain the plasmid. If you want to use it, then you need to add restriction sites and / or expression signals. If you want to to be conjugable then you can add those requisite signals and genes.

But you do not need long stretches of DNA, a plasmid need not be large. My guess is that a few hundred bp is probably the smallest because otherwise you start having steric problems with supercoiling the DNA.

Technically all a plasmid needs is an origin of replication for circular plasmids if it uses the host replication machinery. Linear plasmids also need a terminal resolvase, transposase or telomerase like protein. Depending on the type of origin an additional primase or replicase might also be needed to initiate replication at the origin (eg. m13, geminivirus or circovirus derived plasmids). The selective marker is only to give a fitness advantage to cells with the plasmid over those that lose it so enzymes that cover a nutrient auxotrophy can also be used. Anything else purely depends on what you want the plasmid to do (eg. MCS for further manipulation, secretion systems, protein expression cassettes, metabolic pathways, integration into the genome, transfer to other cells, encapsulation of other plasmids).

As has already been mentioned often the long stretches found on many plasmids are the result of scaring left by restriction site availability. It is still recommended to have a small spacer region between expression cassettes so that the proteins for transcriptional initiation are not sterically hindered (mostly for promoters in opposite directions).