Alternative Splicing can be regulated by:

• Cis Elements, which are elements within the gene's boundaries. The typical example are the Exonic/Intronic Splicing Enhancers/Silencer (ESE/ISE/ESS/ISS), and by by definition these elements are inside the gene coordinates.
• Trans Elements, which represents all the regulators encoded somewhere else in the genome. For example, the Spliceosome, the machinery that carries out splicing, is composed by a variety of small proteins; if one of these proteins is overexpressed (or not expressed), this affects the alternative splicing of all the genes that require it for splicing. If you consider this definition, it means that all events of alternative splicing are regulated by elements outside the gene coordinates.
• There it can be also splicing regulatory elements in the promoter of the gene or in the proximity. Some consider these as in trans elements while other consider them as in cis, which probably creates a bit of confusion.

Your question may be better defined as "are there any alternative splicing cis-regulatory elements outside the gene's boundaries?"

Unfortunately I can't give you any reference on cases of regulatory elements outside the promoters and the gene boundaries, but I believe that there are many. Most of the research to date has focused on studying ESEs and ISEs because they are easier to characterize, but this does not mean that there are not other elements involved. From what we know, splicing may be also regulated by DNA methylation, or other signals. Consider that most of the splicing regulation signals are very degenerated and rely on very few bases for being recognized, so it is probable that other unknown signals are involved.

I am not very sure on the answer and I can not find a reference to link to you, but I would say "Yes there are", unfortunately I can't remind any known case at the moment. Most of the research done has focused on

In biology I'd pretty much never say "all" about any topic...there's always some clever mechanism waiting to be discovered. A lot of what we do is look under the flashlight using techniques and strategies we already understand, leading to finding more of the same thing.

That said, I know there's that cool story of a Drosophila transcript generated by splicing from 2 different strands. That could complicate the machinery.

This paper talks about that:

Alternative trans-splicing of constant and variable exons of a Drosophila axon guidance gene, lola

However, whether all scenarios are real isn't entirely clear I guess. This more recent paper takes on some of the issue:

Global analysis of trans-splicing in Drosophila

I always thought this story was cool, but haven't been following it closely. Nested genes are one of the categories they describe--another complication. So for nested genes their regulatory elements are by definition within other genes. They also mention a few other interesting cases in the discussion.

In terms of your question there would have to be a mechanism that is different from the typical splicing regulatory mechanisms. In general cis-splicing regulatory elements need to be close to the location of the splice-sites for Spliceosome recruitment. If the splicing regulators are far away from the splice-site how could they be involved in the recruitment? I could imagine there would be some kind of loop created to bring them closer to the splice site.

In addition to cis elements present within the pre-mRNA it has been shown that histone modifications can also affect splicing patterns.

EDIT: A snoRNA has also been shown to regulate alternative splicing of a gene found on a different chromosome. However, in the case of the histones and the snoRNA it appears that a cis-regulatory element is required on the pre-mRNA.

Most splicing in eukaryotes takes place within the nucleus. "All" is a tricky word - and I'd have to say "no, all are not found within the gene/transcript" because there always seems to be an exception in biology. Take trans-splicing in C. elegans and other organisms. But that said, don't be concerned with the "all" but with the "most." Do what you can confidently with those tools and data that are reliable.