How much metal needs to be removed?
How far the invisible effect of the crack has extended in to the cymbal is difficult to determine. The only way to be sure would be to subject it to ultrasonic testing, which in most cases would be either impracticable or simply not cost-effective.
Therefore the extent of metal removed is a compromise between ensuring the best chance of removing any stressed but, as yet, not visibly cracked material, and minimising the amount removed so as to affect the sound as little as possible.
Is the crack likely to come back?
As it is not an exact science I can make no guarantee that the crack will not return. However the likelihood of this can obviously be reduced by avoiding striking the repaired area as far as is possible. If it is a large repair, then an angled cymbal will naturally tend to turn on its stand and present the heavier (i.e. unrepaired) edge towards the throne. This will clearly assist in avoiding further damage in the repaired area, but it should be noted that the area being struck will tend to be the same all the time, leading to an increased likelihood of failure on the opposite side to the repair.
Will it sound different?
How much the sound might be affected by the repair is difficult to determine. Obviously a large repair is more likely to change the sound than a small one. Most repairs I have heard back about tend to indicate little or no perceptible change, but some reports have indicated a noticeable difference, to a greater or lesser extent. Of these latter reports, most customers have said that although it sounds a little different they are still happy with the tone, but a very few have said that they were less so.
However most have agreed on two things: firstly, that an unusable cymbal has become usable again, and secondly, that at least the crack has been prevented from spreading to an extent that permanently ruins the cymbal.