A great deal of science is expressed as a form of necessity. But none of it really is, or science itself could not move forward. Is it necessary that when I drop things they fall? Well, yes, but only in a gravity well. In most of the universe, not really. Is it necessary that there is only one line parallel to another through a point? Well, yes, but only at low speeds where geometry is basically Euclidean. And even then, only as an approximation. We may already have found signals that quantum-tunnel faster than the speed of light, at the cost of a certain degree of degradation. Etc.
Such a necessity involves choosing the theories in which one has faith. This is typical of modalities in general.
Basically, modes are always relative. They express that there is an attached context that is presumed to be shared, and therefore is not being expressed. From there you can look at what features a context must have in order to act according to the rules of, say necessity rather than obligation. They do take different forms.
For instance necessary modes allow one to assume their assertions are so seldom violated that they can be ignored. You do not have to plan for their failing.
One does not punish violations of one's physics, one marvels at them and starts over with slightly different physics. When we are less certain of the physics we are using, we actually back off and use the mode of obligation in a backhanded way. This should work... If it fails, there is likely some punishment, but may just be shame on the part of the person who picked the wrong theory or got the computations wrong.
Any time you rule out failure of a given type from explicit consideration, you add contents to the reigning local modality of necessity. Such considerations can range from what is proven mathematically down to what character defects one cannot impute to a judge making a court opinion, or what someone can do and say and still really love you.
From a classical grammatical point of view, this is all metaphorical and there is an underlying real necessity, obligation, desirability, etc. which these continually-changing criteria only mimic in form. But very few people would consider this applicable any more. To think of any kind of absolute meaningful necessity is as big an unjustified leap as imagining an absolute moral law or an absolute standard for what is desirable.
In a universe with absolute necessity, that necessity is logical necessity.
But there is then no need for the notion of necessity. What is logically necessary is simply true, and there is no need for this modality at all.
I think that Wittgenstein's point is exactly that -- that modality itself is really just a kind of sloppiness that does not actually contribute meaning, In reality, there is only really the one modality that is not any way modal, where the 'attached context' is 'truth' and there is exactly one possible world, the world.