what_are_electrons said:
1. Is the electron a cloud in an atom and a point charge when it is free?
In our current understanding (which is of course always subject to revision based on new experimental evidence), the electron is always a point particle, whose position can be specified only by way of a probability distribution (your "cloud"). For an electron bound to an atom, the probability distribution corresponds to the "orbitals" that you can find pictures of in many physics and chemistry textboks. For a free electron, the probablity distribution is a
wave packet that has some width which causes an uncertainty in position at any particular time.
2. If the electron is indeed a cloud when it is inside an atom, then what is the mechanism for the electron to be freed from the atom and to form a free point charge?
Whether inside or outside an atom, an electron's position is always described by a probability "cloud."
3. Is it possible that experimental physicists will one day have the tools to measure the non-point size of the electron?
We already have measured the non-point nature of the electron's probability "cloud" in many situations. All these analyses so far, assume that the electron actually interacts as a point particle (with random position). If the electron did not interact as a point particle, we would have to add "form factors" to the analysis, which are related to the shape and size of the electron. So far this has not been necessary. It
may become necessary in the future, depending on experimental data which is yet to be collected.
For a parallel situation, consider Rutherford's nuclear model of the atom. His analysis of his famous experiments on scattering of alpha particles by gold nuclei assumed that the nuclei could be treated as point particles, and his predicted angular distribution for the scattered alpha particles agreed with his experimental data. Later data using higher-energy alpha particles
disagreed with Rutherford's prediction, and this was taken as evidence that atomic nuclei do in fact have a finite size.
Similarly, when we scatter electrons or neutrinos off of protons or neutrons, at high enough energies the angular distributions of the scattered particles disagree with what we would expect if protons and neutrons were point particles. This was first observed in the late 1960s, and led Feynman to produce his "parton model" in which protons and neutrons were actually collections of "smaller" particles. Further experiments showed that these "partons" in fact had the properties of the "quarks" that had been proposed earlier by Gell-Mann in order to explain the patterns of properties of particles in the burgeoning "particle zoo."
Something similar may very well happen with electrons in the future... but there's been no sign of it yet in experimental data on electron-electron, electron-neutrino and electron-positron scattering, as far as I know.