On the genetics behind caste marriages

Is the son of a couple engaged in farming better suited to the same profession unless he receives a couple of recessive alleles from both parents?

Not everything is coded in the genome, environmental cues play a very important role in shaping who you are. Profession is just one of those things that is not encoded in genes.
Many people do not continue doing the job of their parents, but they do very well in life.
Also note that, aside from the fact that there is no "agriculture" gene, having a recessive allele does not mean you will miss a certain phenotype.
What is true is that someone who has higher education will probably be more keen to have their kids studying. On the contrary, the kids of someone who is a poor farmer and is barely able to provide for his family will probably not have the opportunity of going to school (let alone university) and therefore will continue to work the land. It has nothing to do with genetics.

Generally there is a lot of unscientific information floating around about these matters. One of the grapevines is that it is genetically better for one to marry outside their caste to create diversity(?). Is such a thing true?

Continuing to reproduce in the same group of people will limit diversity in the genome, and create an accumulation of mutations, which is generally a bad thing. That is why it is very common to have particular illnesses in closed communities.

For instance Amish people have an higher incidence of genetic disorders such as Ellis-van Creveld syndrome (a form of dwarfism)

From The Gene for the Ellis–van Creveld Syndrome Is Located on Chromosome 4p16 - Ruiz-Perez et al., Nat. Genetics 2000

Autosomal recessive transmission of the disorder is supported by
multiple affected siblings with unaffected parents in families with
known parental consanguinity. The largest known pedigree segregating
with EVC is that reported by McKusick and his colleagues in the Old
Order Amish (McKusick et al., 1964). The prevalence of living persons
affected with EVC among the Lancaster County Amish was estimated to be
2 in 1000, whereas the frequency of occurrence among live births was 5
in 1000. McKusick et al. estimated that the frequency of heterozygous
carriers in the Old Order Amish is as high as 13% (McKusick et al.,
1964). All cases of EVC in the Amish at that time were traced to a
common founding couple, Samuel King and his wife, who immigrated to
Pennsylvania in 1744.

This is a well known effect, called the founder effect.

As for India, a study on the genetic behind casts has been done a couple of years ago.

Reconstructing Indian Population History - Reich et al., Nature 2010

From the paper:

Haldane wrote 45 years ago that “if inter-caste marriages in India become common, various… recessive characters will become rarer”. However, it has not been generally appreciated that this applies to groups throughout India, and not only to groups in the south where consanguinity is common. We hypothesize that founder effects are responsible for an even higher burden of recessive diseases in India than consanguinity.
To test this hypothesis, we used our data to estimate the probability that two alleles from a group share a common ancestor more recently than that group’s divergence from other Indians, and compared this to the probability that an individual’s two alleles share an ancestor in the last few generations due to consanguinity. Nine of the 15 Indian groups for which we could make this assessment had a higher probability of recessive disease due to founder events than to consanguinity, including all the Indo-European speaking groups.
An additional reason why some diseases are expected to occur at elevated frequencies in India is shared descent from a common Indian ancestral population. An example is a 25 base pair deletion in MYBPC3 that increases heart failure risk by about 7-fold, and occurs at around 4% throughout India but is nearly absent elsewhere.

They conclude that:

We have documented a high level of population substructure in India, and have shown that the model of mixture between two ancestral populations ASI and ANI provides an excellent description of genetic variation in many Indian groups.
[...]
By showing that a large proportion of Indian groups descend from strong founder events, these results highlight the importance of identifying recessive diseases in these groups and mapping causal genes.

It is very important to understand that this results in no way show a genetic basis for the existence of casts, which remain purely social entities. They rather show that an environmental constraint can influence the distribution of genetic traits, due to a restriction in the possibility of mixing genes in the population.