Doubt: Impedance matching

I have a few very basic doubts regarding impedance matching. One or more of them might sound funny to most of you. But if it is so, please clear the doubts.
1. If impedance matching is all about matching the impedance of transmission line with the load, ie an antenna, why is this done only at this junction? Cant there be reflection of signal at other junctions? like from source to transmission line, antenna to the medium (atmosphere). Y dont the waves reflect at these junctions? Y dont we do impedance matching for them?

2. In a practical case, If the signal comes out of an IC, how am I supposed to know the impeadance behind the pin of IC (if not given in datasheet)

Yes, there can be different impedance (or return loss) at antenna connector and the other side of the feeding transmission line. This is the reason that have to keep tracking the impedances all over the transmission path.
Usually the RFIC manufacturers provides the impedance (S parameters) at particular pins of the IC package, and not behind the pin (on the die).

So then we should also bother about impedance matching at the junction between antenna and atmosphere.?

We do. Poor coupling between antenna and its surrounding space results in low radiation efficiency.

In a usual specification, the antenna impedance includes the junction to surrounding space, assuming a standard mounting. Antennas that can be mounted in different ways, e.g. on a conducting plane versus in free space will show a respective input impedance variation. Objects in the antenna vicinity will also affect the impedance.

For example, that is one of the reasons why horn antennas exist. They are impedance transformers, that matches waveguide impedance to the medium impedance...

Essentially, any antenna can be seen as a coupler between an input port and free space. Considering the free space impedance of 377 ohm, some impedance transformation is usually involved.