It is often said that the borders of African countries are almost totally arbitrary, a relic of imperialism. Since Decolonization, has there ever been any instances or serious discussion of borders being redrawn by African governments to better reflect cultural/linguistic/religious boundaries?

First, I'd like to challenge the premise ("the borders of African countries are almost totally arbitrary"). It is a common one, and conceptually it has truth to it in much the same way that all boundaries drawn by people not living in the spaces where the borders pass are arbitrary. Even the concept of a hard, fixed boundary as we use it in legal terms today is a radically different imposition for most African societies, in which people were a far more important resource overall. One only could claim that land where people accepted one's patronage. Again, this is not unlike most of the world before colonialism helped to spread our modern sense of permanent borders.

The challenge to the premise is, however, not about a border being an artificial construct; it has to do with the degree of arbitrariness. European colonial powers were obsessive about getting treaties, angling concessions, 'buying' rights, and claiming protectorates from African authorities. That's part of what was happening in Berlin in 1884-85 (see Katzenellenbogen's "It Didn't Happen in Berlin" in Asiwaju and Nugent, eds., African Boundaries for example): they were trying to determine the rules for activating these claims, and in some cases for resolving overlapping or unclear claims. Hertslet's Map of Africa by Treaty (3 vols. plus maps; Google Books has them) was the English language version of tracking back the validity of claims to the 'original title'--that is, the original cession upon which colonial power was built. Determining the reach of states and rulers, the composition of a 'tribe,' and the courses of rivers, etc., all played a role in settling claims among Europeans.

This doesn't mean that the choices were somehow authentic, [or that] in some cases where nobody could agree or had suitable documents straight lines or swaps couldn't happen. The results were deeply imperfect because they overlaid a fiction; they were mostly arbitrary for being based upon European divination of what the basis of their claims meant. Again, it's still based on an incorrect reading pushed into ill-fitting boxes with the goal of creating the boundary, but it is worth mentioning precisely because the new nationalist orders (mostly urban and European-educated in leadership) that arose in these colonies tended to internalize that fiction and the colony as venue-for-action as a given. The idea that areas are homogeneous, with X area containing one culture, one language, one religion, is also a fiction; I was talking with a colleague a few days ago about his experience in Burkina Faso with highly plural societies by all three of those measures on the village level.

That leads me to the reasons for the paucity of meaningful efforts to undo colonial boundaries: politics. There have been some--usually efforts at erasing rather than adjusting borders--but the major problem is one of inertia. See, at independence, each African nation had a home-grown nationalist elite, or at least one in-waiting, even if it was small (as in the Dem Rep of Congo, for example). For these parties, the colony as it existed was the field for action. Anything that diminished the claim would imperil the political position of that independence party; altering borders might also affect the ability to collect adequate taxes (most retained old colonial modes of tax support and trade policy) and, of course, the transit and communications nets followed the old colonial spaces and lines. It's also worth pointing out that trying to impose such changes by force would invite others to interfere in your affairs, which is one reason the OAU decided at its first meeting to consider the old boundaries set unless and until all parties involved on a particular border agreed to change it. This left independence wars or careful negotiations as the only ways to make such changes. Efforts at the former have rarely succeeded (twice, as far as I can recall).

In that situation, nobody was going to voluntarily climb down from their claims to govern the spaces that had been the nationalist stage for the victory over colonialism--not even in service of broader ideals of unity. When Senegal and Mali briefly federated (for a matter of months) in 1960, this is one reason it collapsed. Tanganyika and Zanzibar became Tanzania with a helpful dose of coercion in 1964, and it's held together mostly because of the overwhelming power of the mainland. Julius Nyerere, by the way, maintained until the end of his life that the only reason they couldn't have made a 'super-bloc' nation in East Africa without borders was because the British trickled independence out so that each country had leaders and ministers who would never surrender power. He was right about the latter, and probably correct that the British (like the French) foresaw local power hierarchies as a barrier to regional unity, but I disagree that it was necessarily intentional.

One must also consider that a major issue is not who a state's boundaries leave out, but who it throws together--and what their history was under colonial rule. Tribalism is dramatically more potent in Kenya than Tanzania, and there are a variety of reasons for that which include a heavy dose of British settler policy.

So although there are cases wherein a state might be more easily or effectively governed if it included all of an additional group, or broke into pieces that were each nominally one group, the reality is that postcolonial politics based on the nation-state as a unit, plus the artificiality of the idea of unitary groups on a local scale, make such efforts politically undesirable (the former) or geographically impossible (the latter). The history of the Ethiopia-Eritrea border shows you just how hard it is to get a negotiated, carefully investigated settlement accepted.

[edit: unclear wording in para 3]