It's worth clarifying that the number of deaths announced per day isn't strictly a measure of who died in the last 24 hours, but instead a measure of how many new deaths have been notified in the last 24 hours. This can include deaths going back days or even a few weeks. I'm fairly sure the data is partially based, for example, on NHS England hospital deaths data -- or at least the sources for these two data sets must be closely related; they cannot be counting different people -- and the NHS data set was seeing updates to deaths in April even during July (albeit only a handful).
As far as I'm concerned, the most reliable data set for deaths due to Covid is the ONS weekly deaths data, which also counts everybody who has died. Although it is always lagging by a couple of weeks, this may actually count somewhat in its favour, since the rush to update daily has also led on occasion to errors. In the two full weeks from July 3rd-July 17th, there was a recorded average (in England and Wales) of around 45-50 Covid deaths daily. As far as I'm aware, this is at worst stable through to today, although we will get a better idea of this in Tuesday's update to the figures I'm citing (first link below). This compares to an average of around 19 deaths daily recorded in English hospitals (second link below), which has fallen in the last two weeks to around 11.
* * * * * *
To answer birdie's question: as far as I'm aware, the number of performed tests has remained stable at around 200k tests daily for a couple of months or so, so any rise in case numbers is likely to be genuine rather than a product of increased testing volume. I'm not sure what the scenario is in hospitals at the moment, but there's good reason to suppose that the rise in cases is currently being driven by younger people, who are apparently much less susceptible to the serious effects of this disease themselves -- although not totally invulnerable, and they can also pass on the disease to others in more danger. I'm not sure this meets either of your last two choices, but it's probably closer to the first one than the second.
https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/datasets/weeklyprovisionalfiguresondeathsregisteredinenglandandwales
https://www.england.nhs.uk/statistics/statistical-work-areas/covid-19-daily-deaths/