The lower bound is not just a BS-specific bound. It is a no-arbitrage bound and so if the price is lower than this, you have an arbitrage opportunity (some good explanation here). It doesn't mean it is present in the market necessarily, because mid price is not necessarily the price you can trade and when you take spread into account this is likely to go away. It is quite often the case for ITM options because data for them is of lower quality (low liquidity).
When the price is exactly on that border (zero time value), it actually would mean that the implied volatility is exactly zero, because you are essentially stating that there is zero probability of price going higher than the strike. Volatility smile is a topic on its own, and there are books written about this phenomenon. From the practical point of view, however, you again need to think about the quality of the data - while in theory you should have some sort of nice smooth volatility smile, when you are working with real data you can get some weird results. First thing is that using ITM options is a bad idea because price quality is more likely bad than good, use only OTM's. Implied volatilities "in theory" should match, but OTM put price is a better estimate of a fair price than ITM call price. Also keep in mind that "smile" doesn't mean that it is symmetric, in fact it can take different kinds of shapes (sometimes quite weird).
Just as a final note, since you are talking about deep ITM calls (or OTM puts), the volatility can go up due to the fact that the prices are quantized. That is, all OTM puts from some point will cost exactly \$0.01, simply because there is nothing below that. Obviously, the implied volatility for higher strike will be higher in such case, with this price fixed and you'll see it going straight up from some point. This can be mistaken for the genuine volatility smile, while it is not it - in theory, option price should go below \$0.01 to fractions and the implied volatility would be completely different.