So is the construction of the Ibrahim Index’s measure of accountability consistent with the concept of legal accountability?

The accountability measure is itself based on six sub-indicators, which are averaged together.

Many of the measures that comprise the accountability index are indeed measures of legal accountability. For example, there is a measure that asks specifically about the existence of legal penalties for officeholders who abuse their positions.

On the other hand, the index also seems to include measures that are less clearly related to legal accountability.

One, which allegedly measures accountability in rural areas, appears to also be a measure of effective decentralisation and participatory planning at local level.

These issues may have something to do with political accountability, but they have little relationship to legal accountability.
Likewise, a separate measure of “accountability in the public sector” includes a measure of civil society access to information on public affairs. Is this really a measure of legal accountability?
There is another problem with the indicators used: Many of them do not ask about accountability, but about corruption.

Is the prevalence of corruption a proxy for a lack of accountability?

It may be true that in countries with very high levels of legal accountability, there is low corruption, and that the reverse is also true. But the index is meant to measure accountability, not corruption.

In any given year, there might be an increase in corruption but also an increase in accountability (e.g., more people go to jail).

The Index might show this as a decrease in accountability if it relies heavily on corruption scores.