I am writing a long paper (around 100 pages). I would consider 50 pages of it interesting in that it solves a problem of some significance in my field and contains an number of difficult ideas in the proof (I think!). The other half is basically case checking using the definitions/lemmas from the first half. The case checks are fairly straightforward but rather tedious and are appendix material. 

My main concern is publishability. I'm hoping that the first half might be publishable at level X but worried the overall length would make it difficult to publish there or maybe almost anywhere.

Thus, assuming things have been done efficiently my question is to ask which of the following options sounds most reasonable, or if there are other options?

1. Submit the 100 page piece to the arxiv and to the journal; let the editor/referee decide to ask the 50 pages to be cut or not. (Concern is that the length leaves an immediate impression of "too long".)

2. Break the 100 pages into a main paper and an appendix. Submit two things to the arxiv
and submit the main paper to the journal with a pointer to the appendix. Figure that if the main paper is acceptable the editor/referee can ask to publish the appendix, or not. If not, the appendix is arxiv only. 

3. Submit the 100 page piece to the arxiv. Then submit the abridged version to the journal, pointing that full details of the case checks are in the arxiv version. (Concern it leaves a bad impression that the submitted version and arxiv version are drastically different.)

Thank you.