Particle accelerators by focused laser beam? My (layman) estimate is that such accelerators could relatively easily and cheap override the current high-end, multimilliard-$ accelerators. Some years ago I've read some interesting breakthrough in the news, but practically nothing since that.
What is the current focus of the research/development? Is there any obstacle found? What is the current estimate to the practical energy limit of such accelerators?
 A: I do not work directly in the field, so I will probably give an incomplete answer, but to my knowledge there are two major obstacles. 
The first consists in keeping the accelerating structure inside the plasma stable. We are not talking about a solid metal radio frequency cavity here, but some charged-fluid nano structures which are very sensitive to a number of perturbations, included the one with themselves.
The second problem is keeping the particle to accelerate within this structures both in the transversal plane (which may come for free as long the plasma bubbles are stable) but also in the longitudinal which is harder since the laser travels the plasma with velocity $<c$ while electrons approach $c$ very rapidly. A different approach consists in using a particle beam instead of a laser beam to excite the plasma, this may sound easier but let me report few lines from the AWAKE project at CERN:

The AWAKE experiment intends to use the existing conventional SPS beam to drive a high amplitude wakefield [- in the plasma] which will then accelerate a witness electron beam. The SPS beam is, however, too long to effectively drive a wakefield. The AWAKE experiment will longitudinally micro-bunch the beam into hundreds of far shorter beams using a wakefield initially driven by a short laser pulse. These beams will then be able to resonantly excite a high amplitude wakefield.

The SPS aka Super Proton Synchrotron is a 7km ring (the one at which W and Z bosons where discovered and that now injects protons in the LHC), so not quite a table-top experiment! And it's just an experiment: something to prove the concept and develop some fundamental technology. It will require much more effort to develop something that can be really used.
This technique is promising but very complex and immature. That's why all the  major new projects of any kind of particle-accelerator machines still rely on RF acceleration.

A 2 GeV electron beam is already pretty interesting: you can have a nice FEL, or, colliding two of them, a $\tau/$charm factory. These machines cost typically few hundreds millions $, they are not at the energy fronter but can produce plenty of physics and eventually Nobel prizes.
Having such a beam over 2cm is for sure very exciting, but the energy is just one of the beam parameters. How many particles are in a bunch, what is the phase space volume occupied by them, what is the repetition rate achievable, how much power is effectively transferred to the beam? These are the kind of questions that you should ask and whose answers still let you go for RF acceleration.
