Yes, you can avoid generating code for that package. But the designers of Go want to avoid looking at the package at all. They have collected data from inside Google which clearly establishes that this is a problem.
You don't have to look into the package to ignore it, I even explained to you how exactly to do it, but you keep repeating this bullshit. You have to look into the package and therefore parse it in one case only: when this package is imported into current namespace, which is almost never the case in Go. You can detect and ignore unused package with looking no farther than current package almost all the time, because each imported package uses unique namespace. If there were no symbols present with that namespace in current package - you don't have to look into the package with that namespace, it's that simple. And gc (Go's compiler) already detects unused packages in similar way.
You do not need to do code genration to establish what the unused packages are. But you must parse and perform semantic analysis of the entire package to determine what the unused packages are. There is more to "compiling" than code genration and optimization.
Parsing is what they want to avoid, not just code generation. Parsing is cheap for small projects. But nothing is cheap at scale. From the talk I have already linked to:
The construction of a single C++ binary at Google can open and read hundreds of individual header files tens of thousands of times. In 2007, build engineers at Google instrumented the compilation of a major Google binary. The file contained about two thousand files that, if simply concatenated together, totaled 4.2 megabytes. By the time the #includes had been expanded, over 8 gigabytes were being delivered to the input of the compiler, a blow-up of 2000 bytes for every C++ source byte.
As another data point, in 2003 Google's build system was moved from a single Makefile to a per-directory design with better-managed, more explicit dependencies. A typical binary shrank about 40% in file size, just from having more accurate dependencies recorded. Even so, the properties of C++ (or C for that matter) make it impractical to verify those dependencies automatically, and today we still do not have an accurate understanding of the dependency requirements of large Google C++ binaries.
If Go allowed including unused packages, you would end up with the exact same problem.